Download The East Asian Miracle: Four Lessons for Development Policy

This PDF is a selection from an out-of-print volume from the National
Bureau of Economic Research
Volume Title: NBER Macroeconomics Annual 1994, Volume 9
Volume Author/Editor: Stanley Fischer and Julio J. Rotemberg, eds.
Volume Publisher: MIT Press
Volume ISBN: 0-262-05172-4
Volume URL: http://www.nber.org/books/fisc94-1
Conference Date: March 11-12, 1994
Publication Date: January 1994
Chapter Title: The East Asian Miracle: Four Lessons for Development
Policy
Chapter Author: John Page
Chapter URL: http://www.nber.org/chapters/c11011
Chapter pages in book: (p. 219 - 282)
JohnPage
THE WORLD BANK
The
East
Four
for
Asian
Miracle:
Lessons
Development Policy
Since the study of economic development began in earnest at the close
of the Second World War, academics and policymakers have debated
the appropriate role of public policy in developing economies. East Asia
has a remarkable record of high and sustained economic growth. From
1965 to 1990 its 23 economies grew faster than those of all other regions.
Most of this achievement is attributable to seemingly miraculous growth
in just eight high performing Asian economies (HPAEs)-Japan; the
"four tigers": Hong Kong, the Republic of Korea, Singapore, and
Taiwan; and the three newly industrializing economies (NIEs) of Southeast Asia, Indonesia, Malaysia, and Thailand.1 The East Asian economies
from Hong Kong's
provide a range of policy frameworks-extending
nearly complete laissez faire to the highly selective policy regimes of
Japan and Korea. The coexistence of activist public policies and rapid
Presented at the National Bureau of Economic Research, Ninth Conference on Macroeconomics. The findings, interpretations, and conclusions expressed in this paper are entirely
those of the author. They do not represent the views of the World Bank, its Executive
Directors, or the countries they represent.
1. These eight HPAEs are the subject of the World Bank's recently completed study, The
East Asian Miracle:EconomicGrowthand Public Policy, on which this essay draws
extensively. The East Asian Miracle report is the product of a World Bank research team
led by John Page and comprising Nancy Birdsall, Ed Campos, W. Max Corden,
Chang-Shik Kim, Howard Pack, Richard Sabot, Joseph E. Stiglitz, and Marilou Uy.
William Easterly, Robert Z. Lawrence, Peter Petri, and Lant Pritchett made major
contributions. Lawrence MacDonald was the principal editor. Background papers are
available from the Policy Research Department, The World Bank. The findings, interpretations and conclusions expressed in this paper are entirely those of the author.
They do not represent the views of the World Bank, its Executive Directors, or the
countries they represent.
220. PAGE
Japan, Korea,
growth in some of the East Asian economies-especially
Singapore, and Taiwan-has raised complex and controversial questions concerning the relationship between government, the private
sector, and the market.
This essay looks at four public policy lessons of the East Asian
miracle. Section 1 argues that the eight HPAEs can be grouped together
and distinguished from other low- and middle-income countries on the
basis of their rapid, sustained, and shared growth. Section 2 examines
the controversy over the sources of growth in the HPAEs and presents
evidence on the relative roles of accumulation and total factor productivity (TFP) change. Section 3 discusses two aspects of public policy in
East Asia that conform to the conventional wisdom concerning good
management and broad-based
development policy-macroeconomic
educational policies. Section 4 examines two more controversial issues
-the significance of the HPAEs' export push strategies and industrial
policies for TFP change. It concludes that export orientation rather than
selective intervention played the dominant role in increasing economywide TFP growth rates.
1. TheNatureof the Miracle-Rapid Growth
with Equity
The HPAEs are a highly diverse group of economies, differing in
natural resources, population, culture, and economic policy. What are
the characteristics that these eight economies shared that cause them to
be grouped together and set apart from other developing economies?
First, they had rapid, sustained growth between 1960 and 1990. This in
itself is unusual among developing economies. Figure 1 shows the
relationship between relative income level in 1960 and per capita
income growth for a sample of 119 countries during the period
1960-1985. This is the standard "convergence picture" first presented in
Romer (1987). The figure also plots an estimated nonlinear relationship
between initial income and growth.2 Per capita income growth is
essentialy independent of the level of relative income in 1960. The fit of
the regression is poor and the significance of individual coefficients
low.3
2. The regression line is the result of a regression of per capita income growth (in
constant 1980 international prices) on 1960 income per capita, including nonlinear
terms up to the second power.
3. Dollar (1991) finds a similarpattern using a sample of 114 countriesand the absolute
level of per capita income in 1960. He finds a clearer pattern in which the lowest
deciles have the lowest per capita income growth rates, the middle-income deciles
have the highest, and the high-income countries are between. He also reports low
significanceof his regressionresults,however.
The East Asian Miracle 221
Figure 1 GDP GROWTHRATE,1960-1985, AND GDP PERCAPITA,1960
0.08
GDP growthrate(average,1960-85)
*an
0.0
0.06 -Korea*
r-
*
HQngKong
SSingapore
*Japan
Thailan
0.04
0.04 IndonI-
lysia
0Indon*
0.022-,
|
[3
--
?
. no
*
i
0
-0.02
ClD
economies
High-incomce
economnues
OtherdevekJping
1
-0.04
0
0.2
0.4
0.6
0.8
1
1.2
Relative per capita GDP, 1960 (percentage of U.S. GDP per capita, 1960)
Note: This figure
GDPG = 0.013
plots this regression
+ 0.062RGDP60
(0.004)
Source: Summers
(0.027)
and Heston (1991);
equation:
0.061RGDP602.
(0.033)
Barro (1989);
N World
119; R2 = 0.036.
Bank
data.
The eight HPAEs are all positive outliers in the income-growth
distribution. While Malaysia, Indonesia, and Thailand are closer to their
predicted values, the remaining five economies, Taiwan, Korea, Japan,
Singapore, and Hong Kong, are all significantly above their predicted
gross domestic product (GDP) per capital growth rates on the basis of
relative income level.4 All of the HPAEs were catching up to the more
developed countries.
Recent work on growth rate persistence suggests that growth rates
for individual economies are highly unstable over time (Easterly et al.,
1993). The HPAEs, however, appear to be an exception. Depending on
the time period selected and the definition of persistent success (in
terms of the fraction of the distribution used to define high growth), the
four tigers plus Japan consistently rank with the handful of persistently
rapidly growing economies. Indeed, Easterly et al. conclude that "the
widespread perception of strong country effects in growth is strongly
influenced by the Gang of Four."
4. Addition of a dummy variable (HPAE = 1) to the estimated equation appearing below
Figure 1 increases the R2 to .242 and greatly improves the precision of the parameter
estimates. The t value on the dummy variable (which is positive) is 4.00.
222 PAGE
AND GROWTHOF GDP, 1965-1989
Figure 2 INCOMEINEQUALITY
An llIo
GDP Growth per Capita (1965-89)
-r
tiXI
Korea*
* Bo swana
Tiwan*
ongKor
** Singapore
0.06
Japan *
Indonesia * Mauritius
iTnisia + ** Egypt
Thailand
0.04
* Gabon
*Malaysia
razil
*Italy
Isr*lAustria
Israe1
-
- Beim^pa
a *Morocco .
-rtLaSrila* Iran ?C
ni,indom
0.02V
..
bi
* Keinya
Paklstan'
ibippiaes
*Venezuela
Malawi *India
angladesho
0
M
-..
.
N A
.
......
I
*Pei
Boivi
^^p
cotedIv
pcotedIvoke
I
I
II
Ghana
Suan ?
*Za*bia
-0.02
-0.04
0I
r
I
I
I4
I
1
10
1
1
I
I
|
I
20
30
IncomeInequality
I
I
I
40
Note: Income inequality is measured by the ratio of the income shares of the richest 20% and
the poorest 20% of the population.
Source: World Bank data.
Where the HPAEs are unique in combining rapid, sustained growth
with highly equal income distributions. The positive association between growth and low inequality in the HPAEs, and the contrast with
other economies, is illustrated in Figure 2. Forty economies are ranked
by the ratio of the income share of the richest fifth of the population to
the income share of the poorest fifth and per capital real GDP growth
during 1965-1989.5
The northwest corner of Figure 2 identifies economies with high
growth (GDP per capita greater than 4.0%) and low relative inequality
(ratio of the income share of the top quintile to that of the bottom
5. Because the timing and frequency of observations on income distribution vary among
countries in the sample, the ratio of the top to bottom quintile is taken at the date
closest to the midpoint.
The East Asian Miracle ?223
quintile less than 10). All of the high-growth, low-inequality economies
are in East Asia. Seven are HPAEs; only Malaysia, which has an index of
of
inequality above 15, is excluded. Within East Asia comparisons
growth rates and Gini coefficients by decade indicate that the distribution of income was substantially more equal in the fastest-growing
HPAEs. Moreover, improvements in income distribution generally coincided with periods of rapid growth (Birdsall and Sabot, 1993).6
2. Fundamentalists,
Mystics,and the Miracle
To observe a miracle is not necessarily to understand it. As with those
attempting to explain religious phenomena, economists analyzing the
"miracle" of East Asia's growth tend to fall into two camps-fundamentalists and mystics.7 Growth fundamentalists
stress the dominant
role of factor accumulation
in explaining the HPAEs' high-income
growth rates. They frequently qualify their explanations by noting that
efficient allocation of resources, once accumulated, must also have been
a part of the East Asian success story-since
such economic laggards as
the countries of the former Soviet Union and Eastern Europe have
accumulated resources at equally impressive rates with radically different results.8
the importance of
Mystics, on the other hand, while acknowledging
accumulation, tend to place greater stress on the role of the acquisition
and mastery of technology. The dominant view-if
there is one-is
that the East Asian economies, particularly Japan, Korea, and Taiwan,
6. Two qualifications should be noted here. First, some studies of Korea have noted
increasing inequality in recent years; however, most of this is due to rising value of
assets, particularly land, rather than increased variation in incomes. Second, reductions
in inequality in Thailand have been relatively minor compared to those in the other
HPAEs, although Thailand's performance is still better than that of most developing
economies.
7. Recently a third group-predestinarians-has
begun to emerge, placing great stress on
the role of initial conditions in 1960, especially education and income equality, in
explaining East Asia's persistently superior growth performance in cross-country regressions. Rodrik (1993) provides a survey of the relevant literature and some crosscountry regressions. We do not consider their arguments further here since, if predestinarian explanations are relevant, there are few lessons for development policy. This is
perhaps why most predestinarian authors ultimately reveal themselves to be fundamentalists or mystics in their policy interpretations of the miracle (see, e.g., Rodrik,
1993).
8. One of the most prominent fundamentalist tractarians, Alwyn Young (1993), has
recently provided a succinct statement of their position. Although he subtitles his essay
a "contrarian view," it is probably the dominant view among the Anglo-American
economics establishment.
224 PAGE
have been unusually successful at catching up technologically to the
advanced economies.9 Empirical support for the mystical position is
adduced from the high rates of total factor productivity growth estimated for industry in several of the HPAEs, notable Japan, Korea, and
Taiwan.
The fundamentalist-mystic distinction is at the heart the debate on
the public policy origins of East Asia's success. Fundamentalists have
stressed East Asia's success in getting the basics right. They tend to
attribute success to policies that increased physical and human capital
per worker and that provided for efficient allocation. They argue that
the successful Asian economies have been better than others at providing a stable macroeconomic environment and a reliable legal framework
to promote domestic and international competition. They also stress
that the orientation of the HPAEs toward international trade and the
absence of price controls and other distortionary policies have led to
and
low relative price distortions. Investments in people-education
health-are important roles for government in the fundamentalist story
(World Bank, 1991).
In contrast to the fundamentalist view, which acknowledges relatively few cases of market failure, mystics contend that markets consistently fail to guide investment to industries that generate the highest
growth. In East Asia, the mystics argue, governments remedied this by
deliberately "getting the prices wrong" to promote industries that
would not otherwise have thrived (Amsden, 1989). The mystics lay
stress on the dynamic gains of activist government policies to alter
industrial structure and promote technological learning, sometimes at
the expense of static allocative efficiency. Moreover, while fundamentalists would explain growth with a standard set of relatively constant
policies, mystics note that the policy mixes used by East Asian economies
were diverse and flexible. They argue that East Asian governments
"governed the market" in critical ways (Wade, 1990). But the crucial
question remains: Have interventions, per se, accelerated growth?
The mystics have provided valuable insights into the history, role,
and extent of East Asian interventions, demonstrating convincingly the
scope of government actions to promote industrial development in
Japan, Korea, and Taiwan. They have successfully shown that East Asia
does not wholly conform to the fundamentalist model. Industrial policy
and interventions in financial markets are not easily reconciled with
9. Amsden (1989, 1993) and Wade (1990) put forward this view. Interestingly, as Young
(1993) points out, so do Balassa (1988) and Krueger (1990). They differ, of course, on the
policy origins of East Asia's presumed success at technological catching up.
TheEastAsianMiracle?225
static allocative efficiency. Policies in some economies are much more in
accord with models of state-led development.
2.1 SOURCESOF THEMIRACLE-ACCUMULATION
AND PRODUCTIVITY
CHANGE
Empirically, the fundamentalist-mystic debate can be viewed in terms
of the relative roles of factor accumulation and productivity growth in
explaining the HPAEs output growth.?0 How unusual are the HPAEs in
terms of their rates of factor accumulation? Are the East Asian economies
atypical in terms of their rates of productivity change?
2.1.1 East Asia's Recordof Accumulation The fundamentalist view of the
success of the HPAEs is that their investment levels in physical and
human capital substantially exceed those for other countries at similar
levels of development, resulting in more rapid growth of per capita
income. Figure 3 shows the relationship between income level in 1960
and the average investment rate for 1960-1985. The estimated nonlinear regression relating initial income level to investment is also shown.
There is substantially more regularity in the relationship between investment share and relative income, than in the relationship between
growth and relative income." The investment rate for all countries
increases with income up to about 70% of U.S. GDP in 1960 and then
declines.
The HPAEs conform much more strongly to the cross-country pattern
of investment rates than to the pattern of growth rates. Thailand and
Hong Kong lie close to their predicted values on the basis of the
cross-country regression. Japan, Korea, and Malaysia are the extreme
outliers among HPAEs, but they are not extreme outliers in the distribution.'2 High investment rates are part of the Asian success story, but
they cannot fully explain the extent to which per capita income growth
in the HPAEs diverges from the typical pattern.
Figure 4(ab) summarizes the pattern of variation of two measures of
human capital with initial income. Figure 4(a) plots the primary school
enrollment rate in 1960 against relative per capita income, while
10. While it is broadly correct to associate fundamentalism with policies to promote
accumulation and efficient, static allocation, some mystics of the industrial policy
school would point to such government initiatives as socialization and bounding of
private risks in Korea and Japan as important policy tools that raised accumulation.
11. The fit of the regression is markedly better, as is the variance of the estimated
coefficients.
12. Addition of a dummy variable (HPAE = 1) to the estimating equation appearing
below Figure 4 does not increase the explanatory power of the regression. The t
statistic on the dummy variable is 1.00.
226- PAGE
Figure 4(b) presents the scatter of Barro and Lee's (1993) average
measure of educational attainment over the period 1960-1985 against
initial income. Both scatters confirm that the HPAEs were relatively well
endowed with human capital but again conform more closely to the
cross-country pattern based on relative income than is the case for
income growth.
2.1.2 Productivity Change TFP is estimated in a simple neoclassical
framework by subtracting from output growth the portion of growth
due to capital accumulation, to human capital accumulation, and to
labor force growth. Assume that every economy has access to an
international cross-economy production function of the form:
Q = AF(K, E, L),
(1)
where A is total factor productivity, K is a measure of capital services,
E is a measure of human capital endowments, and L is a measure of
labor services in natural units.
TFP change can be then found as the residual of growth of output
per worker after deducting the contributions of human and physical
OF GDP,
Figure 3 AVERAGEINVESTMENTRATEAS A PERCENTAGE
1960-1985, AND GDP PERCAPITALLEVEL,1960
40
Averageinvestmentrate(percentageof GDP,1960-85)
*
.
*Japan
*aysia
l--*.
r.aysg-ia*
Singapore
30 -
*
20......- ...........
20
.*.o_
e
Oe
10
g
-
0
Kong
-
0 *e
'.
*.::
10
9
E
.
?
^ *
*DHPA-s
High-incone economies
* Otherdeveiopingeconomies
*
0*
0
0
E
;
t
a[0
D
*
?
Korea ^e *
* *'
* *
Taiwan*
?
0.2
0.4
0.6
i
0.8
Relativeper capitaGDP (percentageof U.S.GDP percapita,1960)
Note: Regression equation:
16085 = 10.125 + 59.120RGDP60 - 51.881RGDP602. N = 119; R2 = 0.295.
(12.593)
(1.383)
(10.344)
Sources: Summers and Heston (1991); Barro (1989); World Bank data.
1
1.2
The East Asian Miracle 227
Figure 4a INITIAL INCOME LEVEL AND PRIMARY ENROLLMENT RATE
PRIM6O
1.6
1.4tl
1.2 r
.
Sin lve
t
-
Korea
0.8
SI *
Inconsisa
Hong Kong
,y
""
Thaiiland*~
/
.
0.6-
I
0.2
-
*
0
0
mt ?
*
t
t
t
!
i
I
I
!
I
i
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
1.1
PER CAPITAGDP 1960 (% USA GDP/CAP 1960)
Figure 4b INITIAL INCOME LEVEL AND AVERAGE EDUCATION STOCK
12
AV. EDUCATION
(60-85)
0
0
0
00
0
10-
0
1*Japan
8 -
0
0. .0
a
0.1
/
0.2
0.3
0.4
0.5
0.6
0.7
0.8
PER CAPITAGDP 1960 (% USA GDP/CAP 1960)
0.9
1
1.1
228 PAGE
capital accumulation:
a = (q - 1) - sK(k -
) - sE(e-
1).
(2)
Under assumptions of competitive factor markets and constant returns
to scale, SK and SE are equal to the income shares of factors. Thus, most
empirical applications of Equation (2) estimate the output elasticity
coefficients with income shares.13
Because income share data are not available for most countries in our
sample, we instead estimate the output elasticities directly using a
simple, cross-economy production function. We regress annual log
output growth on log capital growth, log human capital growth, and
log labor growth between 1960 and 1990, constraining their coefficients
to sum to unity (i.e., specifying the production function to be
Cobb-Douglas). We also include economy-specific dummy variables to
estimate individual rates of TFP change for each of the sample's
economies.
We estimate TFP change for 87 high- to low-income economies. The
data set includes new constant price capital stock data (Nehru and
Dhareshwar, 1993). Measures of human capital are incorporated in the
specification using Barro and Lee's (1993) measure of educational attainment. Table 1 reports the production function parameters and the
estimated TFP growth rates.
The low elasticity of output with respect to capital from the crosscountry sample is striking but not altogether surprising. There is a
subset of developing economies in our data (13 in number) that have
positive net investment and human capital growth per worker, but
negative output per worker growth rates. In effect, the marginal product of both physical and human capital was negative in these economies,
an indication of severe allocation inefficiency. Empirically, this is reflected in a lower elasticity of output of both physical and human
capital in the production function based on the whole sample than
when only economies with nonnegative growth of output per worker
are included.
We have reestimated the production function based only on highincome economy input-output relationships. This is predicated on the
assumption that allocation efficiency in the high-income economies is
greater than in the whole sample and, hence, that TFP growth rates
estimated on the basis of the production function parameters will
contain loss "noise" due to allocation mistakes. These results and the
estimated TFP growth rates derived from standard growth accounting
13. This literature is briefly surveyed in Nishimizu and Page (1987).
TheEastAsianMiracle?229
methods are displayed in Table 1. The elasticity of output with respect
to capital in the high-income economy production function rises to
more conventional levels. TFP estimates, particularly for those
economies within rapidly growing capital stocks, are correspondingly
reduced.
Figure 5 compares the TFP growth estimates from the two production
functions. Three inferences are supported by both sets of estimates:
1. The range of TFP growth rates for high-income countries is quite
compact, especially in comparison with the low- and middle-income
countries.
2. Nearly one third (32%) of the low- and middle-income countries in
the sample had negative rates of TFP growth for the period
1960-1989, regardless of the parameter estimates used.
3. There is very little productivity based "catch-up" exhibited by the
low- and middle-income countries.
Table1 ELASTICITY
OF OUTPUTWITHRESPECTTO CAPITAL(SK),
LABOR(SL),AND HUMANCAPITAL(SH):FULLSAMPLE
AND HIGH-INCOME
ECONOMIES
Observations SK
Full sample
High-income
economies
2,093
460
(t-stat)
0.178
0.399
SL
10.895 0.669
10.237 0.332
(t-stat)
SH
(t-stat)
6.411
1.679
0.154
0.269
1.49
1.476
RESULTING
TOTALFACTORPRODUCTIVITY
GROWTHESTIMATES
FORTHEHPAES(1960-1989)
Economy
TFPgrowth
(full-sample
parameter
estimates
Hong Kong
3.6470
Indonesia
1.2543
Japan
Korea,Rep. of
3.4776
3.1021
Malaysia
Singapore
Taiwan
Thailand
Latin America
Africa
Source:World Bank data.
TFPgrowth(highincomeonly,
parameter
estimates)
2.4113
- 0.7953
1.4274
0.2355
1.0755
-1.3369
1.1911
3.7604
2.4960
-3.0112
1.2829
0.5466
0.1274
-0.9978
-0.9819
-3.0140
230- PAGE
Figure 5 COMPARISON OF TOTAL FACTOR PRODUCTIVITY ESTIMATES
TFP growth (high-income
sample parameters),
1960-89,
percent
4
2
0
-2
-4
-6
-4
-3
-2
-1
0
TFP growth (full sample parameters),
1
1980-89,
2
3
4
percent
These are not promising results for the developing world. Despite a
substantial literature on the potential for developing countries to achieve
rapid growth through the adoption of known, "best practice" technologies, very few countries appear to have realized these potential gains.14
Catch-up, where it is taking place, is due primarily to higher rates of
factor accumulation.
Three of the HPAEs are in the upper decile of the TFP distribution in
both sets of estimates-Hong
Kong, Taiwan, and Japan. Two othersThailand and Korea-are in the upper decile of the distribution based
on the full sample production function parameters, but they are unremarkable on the basis of the high-income country production function
parameters.l5 Three-Indonesia,
Malaysia, and Singapore-shift from
14. For a concise review of the arguments for technologicallybased catch-up, see Pack
(1993).
15. When we compare our estimates of TFP growth with two other independently
derived estimates for a large sample of countries (Elias, 1991; Fischer, 1993), the
pattern of productivitygrowth rates in Figure5 is remarkablyrobust to the specification of the growth accounting equation and to the capital stock series used. The IMF
The East Asian Miracle *231
modest, but positive, TFP growth to negative TFP growth when the
production function parameters are changed from those estimated from
the whole sample to those derived from the high-income sample.
Presumably in the high-income countries, most of the estimated TFP
growth is due to technical progress-advances in best practice-which
explains the relatively compact distribution of TFP growth rates between 0.5% and 1.5% per year, and the tendency among the highincome countries for TFP growth to decline with rising income (World
Bank, 1993). In low- and middle-income countries, however, changes in
TFP must reflect more than technical progress, otherwise we would
never find negative TFP growth rates.
We have already argued that TFP growth for low- and middleincome economies contains an element of catching up to (or falling
behind) best practice technologies. But TFP growth rates in a one-sector, cross-country estimate will also contain an element of allocative
efficiency; economies that allocate physical and human capital to low
yielding investments will have low or negative estimated TFP growth
rates. The clearest demonstrations of this are the 13 economies with
negative output growth and positive accumulation.
In an attempt to look at patterns of technologically based catch-up,
we assume that the elasticities of output that should be used to
calculate TFP change are those that apply to the high-income economies
only. We subtract the average rate of TFP change for the high-income
economies, which we associate with movements in international best
practice, from TFP change to get an estimate of technical efficiency
change. Using this method Hong Kong (2.0%), Japan (1.0%), Taiwan
(0.8%), and Thailand (0.1%) are the only HPAEs catching up to international best practice. Korea (-0.2%) was essentially just keeping pace
with technological progress in the high-income economies, while the
investment driven economies of Indonesia, Malaysia, and Singapore
were falling behind international best practice at rates of 1.2% (Indonesia) to 3.5% (Singapore) per year (Table 2).16
(1993) WorldEconomicOutlook,contrastingAsia with other developing economies,
reaches similar conclusions, both with respect to the estimated magnitudes of TFP
change and the relative contributionof TFP change to output growth. Thomas and
Wang (1993) and Edwards (1992) also reach broadly similar results concerning the
pattern of productivitychange in the HPAEscomparedwith other economies.
16. Friedberg,Khamis,and Page (1993),using a cross-countrystochasticfrontierproduction function to estimate movements in international best practice, reach similar
conclusions with respect to the pattern of technical efficiency change among the
HPAEs.
232 PAGE
Table2 TECHNICAL
EFFICIENCY
CHANGE
ESTIMATES
FORTHEHPAES
Economy/ Region
Hong Kong
Indonesia
Japan
Korea, Rep. of
Malaysia
Technicalefficiencychange,
1960-1989
1.9714
-1.2352
0.9876
- 0.2044
-1.7767
Singapore
- 3.4510
Latin America
-1.4217
Africa
- 3.4539
Taiwan
Thailand
0.8431
0.1067
Source:World Bank data.
When the HPAEs are contrasted with other developing regions,
however, their ability to keep pace with international best practice
seems somewhat more remarkable. Using the same method, we have
estimated the average rate of technical efficiency change for Latin
America (-1.4%) and Africa (-3.5%). Against these benchmarks, all of
the HPAEs except Singapore stand up well in terms of their ability to
keep pace with the world's shifting technological frontier.17
What, then, are we to make of the fundamentalist-mystic debate?
Figure 6 suggests that each camp can claim partial victory. The diagram
shows the relative contribution to output growth of factor accumulation
(share-weighted total input growth) and TFP growth.
When the TFP estimates based on the high-income country sample
are used, the dichotomy between investment driven and productivity
driven economies is dramatic.18 Rapid growth in Indonesia, Malaysia,
17. Alwyn Young (1992) in a well-known paper finds similarly disappointing results with
respect to Singapore's TFP performance. Some caution is needed in interpreting these
results, however. Much of Singapore's investment between 1960 and 1990 was in
housing and social infrastructure, output of which are notoriously difficult to measure. It is possible that we have undervalued the rate of growth of output and, hence,
the rate of TFP change. Similar detailed criticisms could be made for other economies,
both HPAE and non-HPAE, and our TFP results are best regarded as indicative of
broad international trends.
18. Despite the low elasticities of output with respect to capital in the full sample
production function, only 10 of the 60 non-HPAE low- and middle-income economies
have contributions of TFP growth exceeding 30%. Among the HPAEs, the
to a
investment-driven economies-Indonesia,
Malaysia, and Singapore-conform
typical developing economy pattern with a low (but positive) TFP contribution to
output growth. The productivity-driven economies-Japan, Korea, Taiwan, China,
Hong Kong, and Thailand-on the other hand are more unusual and look more like
advanced economies with a relatively high contribution of TFP to output growth of
above 30%.
The East Asian Miracle ?233
and Singapore is wholly due to high rates of factor accumulation.
Indeed, their productivity performance, like that of the vast majority of
developing economies, detracted from the potential output growth that
could have been achieved, based on their rates of factor accumulation.
Hong Kong, Taiwan, and Japan in contrast had relatively high TFP
growth rates and derived from 15% to 30% of their output growth from
TFP growth, a characteristic shared by only four other developing
economies. Thailand and Korea had rates and relative contributions of
were
TFP growth that were atypical for developing economies-they
for
The
than
the
relative
share
advanced
economies.
less
positive-but
and
of
of
sources
Korea,
Hong
Kong,
growth
Singapore, all
contrasting
of which shared essentially the same rate of growth, are a good
illustration of the varying roads to rapid growth among the HPAEs.
The East Asian success story is, then, primarily a fundamentalist one.
Depending on the estimates used, between 60% and 120% of their
output growth derives from accumulation of physical and human
capital and labor force growth. The results also suggest that the HPAEs
were unusually successful in allocating these factors of production. The
estimates of TFP growth based on the full sample address the following
question: Based on the average efficiency with which physical and
GROWTHAND PARTOF
Figure6 TOTALFACTORPRODUCTIVITY
GROWTHDUE TO GROWTHOF FACTORINPUTS,1960-1989
(HIGH-INCOMESAMPLEPARAMETERS)
4
TFP GROWTH
(percent p.a.)
30%
20%,
2
2* wan
Tai
C Japa
*1
-4
-2
0
2
4
6
Total Factor Input Growth (percent p.a.)
8
~
10
12
234 PAGE
human capital are used in the world economy, does accumulation overor underpredict income growth? The answer is that for most low- and
middle-income economies, it overpredicts growth, while for the HPAEs
it underpredicts. Even when the production function is estimated from
the high-income sample only, accumulation underpredicts growth for
five of the eight.
Nevertheless, mystics can claim some victories. The subset of productivity-driven HPAEs are unusual among developing economies because
of the relatively important role of TFP in their growth, even when
attempts are made, as here, to correct for the gains from superior
accumulation.9 They are also among the very few developing countries
keeping pace with or catching up to the moving target of international
and
best practice. Two of the productivity-driven economies-Taiwan
with
identified
economies
most
the
East
Asian
Japan-are among
activist policies to promote technological learning. Hong Kong-the
most productivity driven of all of the HPAEs-in contrast is the least
interventionist of the group. Taiwan, Korea, and Japan are economies in
which activist policies to allocate resources were undertaken, yet apparently without the deleterious effects that characterized the performance
of other economies with similar policies.
2.2 THEPOLICYORIGINSOF SUCCESS
The HPAEs have achieved rapid growth through successful attainment
of both fundamentalist and mystic objectives. They have performed
better than most low- and middle-income countries three critical funcallocation, and technological catch-up. They did
tions-accumulation,
this with combinations of policies ranging from market-oriented to state
led that varied both across economies and over time.
Despite the diversity there are a number of common policy threads
that bind the HPAEs. Macroeconomic management was unusually good
and macroeconomic performance unusually stable, providing the essential framework for private investment. Policies to increase the integrity
of the banking system, and to make it more accessible to nontraditional
savers, increased the levels of financial savings. Education policies that
focused on primary and secondary education generated rapid increases
in labor force skills. Agricultural policies stressed productivity change
toward best practice-should be free of the
19. Technical efficiency change-movement
allocative gains characteristic of TFP estimates by growth accounting methods (see
Nishimizu and Page, 1982). Friedberg, Khamis, and Page (1993), using a stochastic
frontier production function, find that the HPAEs are the only regional grouping of
developing countries that have consistently closed the gap with international best
practice.
The East Asian Miracle *235
and did not tax the rural economy excessively. All of the HPAEs kept
price distortions within reasonable bounds and were open to foreign
ideas and technology.
But these fundamental policies do not tell the entire story. In most of
these economies, in one form or another, the government intervened
and through multiple channels-to
foster develop-systematically
ment, and in some cases the development of specific industries. Policy
interventions took many forms-targeted and subsidized credit to selected industries, low deposit rates and ceilings on borrowing rates to
increase profits and retained earnings, protection of domestic import
substitutes, subsidies to declining industries, the establishment and
financial support of government banks, public investments in applied
research, firm- and industry-specific export targets, development of
export marketing institutions, and wide sharing of information between
public and private sectors. Some industries were promoted, while
others were not. These strategies of selective promotion were closely
associated with high rates of private investment and, in some of the
HPAEs, high rates of productivity growth.
The relative importance of policy fundamentals and selective interventions has been and will continue to be at the center of the debate
over the policy origins of East Asia's success. To consider in any depth
the catalogue of policies outlined earlier exceeds the scope of this essay.
Rather, we examine four possible lessons for public policy arising from
the HPAEs success.
3. ConventionalWisdom-Macroeconomic
Management
and EducationalStrategy
Among fundamental economic policies that have contributed to East
Asia's superior record of accumulation and allocation of resources, two
stand out-good macroeconomic management and emphasis on broadly
based education.20 The nature of the policies pursued and their impact
on growth in the HPAEs are described briefly in the following.
3.1 MAINTAININGMACROECONOMIC
STABILITY
In contrast with many other developing economies, the HPAEs have
been remarkably successful in creating and sustaining macroeconomic
20. World Bank (1993) enumeratesa number of other "policy fundamentals"that have
acted primarilyto increase accumulationand improve allocation.These are omitted
here to allow for a fuller discussion of the two described.Of the omitted fundamentals, perhaps the most significant is the focus on building secure, bank-based financial
systems.
236 PAGE
Table3 CONSOLIDATED
PUBLICSECTORDEFICITS,
SELECTED
EASTASIANAND OTHERECONOMIES
Economy/ Region
Averagepublic deficit,
percentageof GDP,
1980-1988
Rankamong 40
developingcountries
(1 = highest deficit)
1.89
10.80
5.80
6.39
2.82
34
6
23
HPAEs
Korea,Rep. of
Malaysia
Thailand
Average,40 developing economies
Average,OECDeconomies
Other economies
Argentina
Brazil
Mexico
Philippines
9.62
4.02
6.73
4.40
Source:Easterly, Rodriguez, and Schmidt-Hebbel (forthcoming).
stability. Here we consider the HPAEs' successful management of three
macroeconomic variables: budget deficits, external debt, and exchange
rates.
3.1.1 Budget Deficits and Inflation International experience suggests that
the macroeconomic consequences of public sector deficits depends on
how they are financed. Although the HPAEs' budget deficits are not
dramatically smaller as a group than those of other developing
economies, they were better at keeping deficits within the limits imposed by their ability to finance them without destabilizing the macroeconomy.21 The financing limits themselves were also higher due to the
more rapid growth of the HPAEs.
Table 3 shows consolidated public sector deficits for the 1980s for
three developing country HPAEs that have good data compared with a
sample of OECD and other developing economies. As a percentage of
21. The analysis in this section draws on W. Easterly and K. Schmidt-Hebbel, "The
Macroeconomics of Public Sector Deficits: a Synthesis" in W. Easterly, C. Rodriguez,
and K. Schmidt-Hebbel, eds., Public Sector Deficits and MacroeconomicPerformance,
Oxford University Press (forthcoming). The data on consolidated public deficits, as
well as the rest of the data in this section except where otherwise indicated, are from
the same source. Consolidated public deficits, though less widely available than
central government deficits, are a much more reliable indicator of fiscal management,
because they include operating deficits of public enterprises that have played a critical
role in some macroeconomic crises.
TheEastAsianMiracle?237
GDP, Korea's budget deficits were below even the OECD average.
Malaysia and Thailand are more complicated. Thailand's budget deficits
were about average for developing economies in the 1980s, while
deficits in Malaysia were substantially bigger than average. Both ran
larger budget deficits than such troubled economies as the Philippines,
Brazil, Argentina, and Mexico.
Unlike these and other economies that encountered difficulties, however, Malaysia and Thailand successfully financed their deficits. This
was possible for the following reasons:
* First, there was feedback from high growth. Because rapid growth
increased the demand for financial assets, Malaysia and Thailand
were able to absorb higher levels of monetary financing without a
rapid rise in inflation. Moreover, rapid growth in GDP raised the level
of sustainable domestic and external borrowing.
* Second, there was feedbackfrom high financial savings. Savings rates
were high in Malaysia and Thailand, and much of this saving went
into the domestic financial system (as opposed to real assets or capital
flight as in Latin America). This further increased the demand for
money and other domestic financial assets, making increased domestic financing of the deficit possible without resort to inflationary
financing. In Malaysia, the government Provident Fund mobilized
domestic saving for the government's use in noninflationary financing of the deficit.
* Third, there were low initial debt ratios. In Thailand, the initial level of
external debt to GDP was very low, which meant that external
financing was available when needed.
By holding public deficits within the bounds of prudent financing,
the HPAEs have avoided the inflation-inducing bursts of money creation that afflict other developing economies. Figure 7 shows money
creation as a ratio to GDP in Korea, Malaysia, and Thailand and in three
unstable comparators, Argentina, Mexico, and Zaire. The contrast is
striking: While money creation has been relatively constant among the
HPAEs, each of the comparators experienced two episodes of rapid
money creation when fiscal balances deteriorated or external financing
dried up.
Moderate inflation was a corollary of fiscal prudence (Table 4) .22 East
Asian governments never had to rely heavily on the inflation tax
22. Recentresearchsuggests that inflationbelow 20%,a level not breachedby any of the
HPAEs during their rapid growth periods, can be maintained for long periods
without generatingmacroeconomicinstability(Dombusch and Fischer,1993).
238 PAGE
because their deficits were within financable limits. In general, HPAE
governments have been strong enough to alter public spending and
foreign borrowing as needed, although in Thailand this has been a
continuous struggle (Warr and Nadhiprabha, 1993).
One result of low-to-moderate inflation rates particularly welcome to
business is stable real interest rates. Figure 8 shows real interest rates in
Korea, Malaysia, Thailand, Argentina, Ghana, and Mexico. As with
money creation, the contrast is remarkable. In the East Asian cases, low
inflation and flexible financial policies kept real interest rates within a
narrow range.
OF
Figure 7 REVENUESFROMMONEY CREATIONAS A PERCENTAGE
GDP:EXAMPLES
FROMEASTASIAAND OTHERSELECTED
ECONOMIES
14
(Percent)
.
12
Rep,of Kora_...
-Thailand
--.....
10
tsii
6
-
---
i
!
4
2
0
igII
\n---I
t
-2
iI
i
i
1970197219741976197819801982198419861988
14
1-^
I_pi
I-?--t-t
g
_____
___
/\
--TA~ire
{
j__
I
2
o
A
1970 1972 1974 1976 1978 1980 1982 1984 1986
Note: Revenues from money creation as a percentage
change in high-powered money to nominal GDP.
Source: World Bank data.
of GDP is defined as ratio of nominal
TheEastAsianMiracle?239
3.1.2 Managing Foreign Debt Of the seven developing HPAEs, only
Indonesia, Korea, Malaysia, and Thailand have public or publicly guaranteed foreign debt. The governments of the others-Singapore, Hong
Kong, and Taiwan-have not borrowed abroad. None of the four with
foreign debt has faced a crisis in the sense of having to reschedule debt,
but sharp increases in debt have led to rapid adjustment.
In Korea in 1980-1985, Malaysia in 1982-1988, and Indonesia since
1987, debt to GNP ratios have been quite high compared with other
indebted economies (Table 5). As with fiscal deficits, however, favorable
feedback from other policies enabled the HPAE debtors to sustain
higher ratios of external debt to GDP than other economies. In particular, high levels of exports have meant that foreign exchange was readily
available to service the foreign debt. For example, when Mexico faced
severe problems with its creditors in 1982, it had a much lower debt to
GNP ratio than Korea in 1984 but a much higher debt to export ratio.
3.1.3 ExchangeRate Management The evolution of exchange rate regimes
in the HPAEs has been broadly similar. Most moved from long-term
fixed rate regimes to fixed-but-adjustable rate regimes with occasional
steep devaluations to managed floating rate regimes. Due to the combination of moderate inflation and active exchange rate management, the
HPAEs avoided the severe exchange rate appreciation that beset Africa
Table4 INFLATIONRATES
/ Region
Economy
AverageCPI,
1961-1991
HPAEsa
7.5
Hong Kongb
8.8
Indonesiac
Korea,Rep. of
Malaysia
Singapore
Taiwan
Thailand
All low- and middle-incomeeconomies
South Asia
Sub-SaharanAfrica
LatinAmericaand Caribbean
12.4
12.2
3.4
3.6
6.2
5.6
61.8
8.0
20.0
192.1
aAverages are unweighted
b1972-91 only.
c1969-91 only.
Sources: World Bank data; World Bank (1992); Taiwan (1992).
240- PAGE
Figure 8 REAL INTEREST RATES:EXAMPLESFROM EAST ASIA AND OTHER
SELECTED ECONOMIES
20
Real Interest Rate (percent)
rCi
O
I
T
-20
,,/
-30
70,
-80
r--..
,
---
o or
r
alysiw'-.
T
I
-90
ffd----------.-i
-100
1970
20/
1972
1974
1976
1978
1980
1984
1982
1986
Real Interest Rate (percent)
i
V
i
-40
-50
-60
t
1
~~~~i-70
--ArPgeniar
;hn
-80
-90
i
-100
I
1970
1972
1974
1976
1978
M
co
I
1980
1982
1984
1986
Note: Real interest rates are defined as the deposit rate deflated by the consumer price index.
Bank data.
data.
Source:
Source:World
World Bank
TheEastAsianMiracle*241
and Latin America (Table 6) and achieved unusual stability of the real
exchange rate. The HPAEs' success at maintaining stable real exchange
rates is apparent in Figure 9, which contrasts real exchange rates since
1970 in Korea, Malaysia, and Thailand with those in Sri Lanka, Argentina, and Peru.
RESPONDINGEFFECTIVELY
SHOCKS One important
TO MACROECONOMIC
element of the HPAEs' capacity to maintain macroeconomic stability
has been their prompt and effective response to macroeconomic shocks.
The HPAEs made definite policy decisions to keep the macroeconomy
under control, rather than simply benefiting from a feedback from rapid
growth to macroeconomic stability. These decisions are illustrated by
the experiences of Korea and Thailand in adjusting to rather different
macroeconomic shocks.
KOREA
RETRENCHES
In 1979, Korea encountered a variety of problems
that threatened to undercut the 1970s impressive growth. Rising oil
prices depressed the terms of trade, the world recession dampened
export demand, and high interest rates increased debt service costs.
Real appreciation during the 1974-1979 fixed exchange rate regime had
made exports less competitive, the rice crop had failed, and the assassination of President Park Chung-Hee had exacerbated political uncertainty (Collins and Park, 1989). Korea responded with an aggressive
1980 stabilization package backed by IMF standby credits. The governTable 5 INTERNATIONAL INDEBTEDNESS
Ratioof totaldebt
to GNP
Economy/ Region
Peak yeara
Ratioof totaldebtto
exported
goods
andservices
1991
Peak yeara
1991
225.6
HPAEs
Indonesia
69
66.4
263.5
Korea,Rep. of
52.5
15.0
142.4
86.5
47.6
138.4
54.2
47.8
39.0
171.7
94.8
Malaysia
Thailand
All low- and middle-income
economies
South Asia
Sub-Saharan Africa
LatinAmericaand Caribbean
45.2
38.4
176.2
29.6
106.1
293.3
340.8
37.4
268.0
a1987 for Indonesia, and 1985 or 1986 for the other three countries.
Source:World Bank data.
242 PAGE
ment ended the fixed exchange rate regime, devalued the won by 17%,
and tightened monetary and fiscal policy. In 1980, output fell 5%,
inflation exceeded 25%, and the current account deficit approached 9%
of GDP. By 1982 inflation dropped to 7% and in 1983 to 3.4%. The
current account deficit fell to 2% of GDP in 1983.
IN THAILANDThailand only partially adjusted to
GRADUAL
ADJUSTMENT
the first oil shock and in the late 1970s engaged in a mild private and
public spending boom. By 1980-1981, the consolidated public sector
deficit was 7% of GDP, nearly half of which was the deficit of the
nonfinancial public enterprises, and the current account deficit was also
about 7%. Because past foreign borrowing had been moderate-the
debt/GDP ratio was only 35% in 1982-Thailand continued to have
access to international capital markets. Even so, the new government
that took over in 1980 perceived that macroeconomic adjustment was
needed.
Monetary policy options were limited by the fixed exchange rate and
the relatively open capital market. Therefore, the government took the
alternative path, fiscal contraction, moving gradually but consistently
Table6 AVERAGEAPPRECIATION
INDEX,1976-1985
Economy
HPAEs
Hong Kong
Index(highervalue
meansmore
appreciated)a
rank(100
Percentage
meansmost
0 least)
appreciated,
64
1
98
25
Korea,Rep. of
Malaysia
Singapore
Taiwan
Thailand
110
88
87
116
75
41
12
11
47
5
Otherselectedeconomies
Argentina
Bolivia
113
181
45
89
Cote d'Ivoire
185
90
Ghana
248
99
Nigeria
Zaire
277
201
100
95
Indonesia
aDollar's index is based on Summers-Heston purchasing power parity comparisons. An index
value of 100 signifies that the economy's deviation from PPP is where it should be given its per
capita income.
Source: Dollar (1992).
The East Asian Miracle 243
Figure 9 EXAMPLES OF REAL EXCHANGE RATE VARIABILITYIN EAST
ASIA AND OTHER SELECTED ECONOMIES
350
Real E-xchangze
Rate (percent)
300
~~
~
~
~
otKorei
i-Rep
200
150
0
1970 1972 1974 1976 1978 1980 1982 1984 1986 1988
350
Real ExchangeRate (percent)
300
L
L
--Argentina
~~~~~-Mexidco
-?eru
150
100
.......
0
1970 1972 1974 1976 1978 1980 1982 1984 1986 1988
Note: Index of real exchange rate: 1980
Source:World Bank data.
100; real depreciation
is down.
244. PAGE
over the next several years to cut expenditures and increase revenues.
Policymakers steeply cut deficits of the nonfinancial public enterprises,
then gradually reduced the central government deficit. As a result, the
consolidated government deficit declined from 8% of GDP in 1981-1982
to 1.6% in 1986-1987, when adjustment was essentially complete.
Meanwhile, steeper tax rates and tougher collection efforts boosted
central government tax revenue from 13% of GDP in 1982 to 16% in
1988.
3.1.4 MacroeconomicManagementand Rapid Growth It cannot be a coincidence that all of these seven economies have been exceptional highgrowth economies by world standards, and all have had unusual
success managing their macro economies over the long run. Cross-economy, econometric studies generally find that higher inflation, larger
budget deficits, and distorted foreign exchange markets reduce growth
(Fischer, 1993; Rodrick, 1993). All of the HPAEs but Indonesia and
Korea have been long-period, low-inflation economies, while Indonesia
and Korea fall into the moderately low-inflation category.
The channels by which inflation and budget deficits reduce growth
are also of some interest. Fischer (1993) concludes that high inflation
and high-budget deficits reduce both the rate of capital accumulation
and the rate of productivity change. Uncertainty, arising primarily from
the variability of inflation and the inconsistency of relative price signals,
reduces both private investment and the efficiency of resource allocation.
The HPAEs provide at least partial confirmation of this hypothesis.
Figure 10 shows private and public investment as a share of GDP for a
sample of 47 low- and middle-income countries (including five of the
developing country HPAEs) for which consistent data are available.23
The HPAEs are remarkable for their high share of private investment.
Private investment is about seven percentage points higher in the
HPAEs than in other middle-income economies. It rose from about 15%
of GDP in 1970 to nearly 22% in 1974, then declined and held at about
18% between 1975 and 1984. Private investment contracted sharply
between 1984 and 1986, reflecting the global recession, then recovered
by 1988.24 In contrast, private investment in other low- and middle-income countries has remained relatively stable at about 11% of GDP.
23. The data are drawn from Pfeffermann and Madarassy (1992).
24. This basic pattern is observed in four individual economies-Korea, Thailand, Singapore, and Malaysia. The pattern for Indonesia differs; real private investment declined
continuously during the 1980s from a peak of 20% of GDP to a low of 13% in 1989.
The East Asian Miracle ?245
One important element of the HPAEs' capacity to maintain macroeconomic stability has been their prompt and effective response to macroeconomic shocks. This has been greatly facilitated by two characteristics.
First, by limiting transfers to public enterprises and tightly supervising
banks, governments reduced the spillover from the real sector into the
financial sector that in other economies exacerbated fiscal woes. Second,
flexible labor and capital markets enabled the real sector to react quickly
to government initiatives, setting off new growth cycles that eased the
recessionary impact of stabilization measures.
Effective responses to macroeconomic shocks may also have contributed to the HPAEs' long-run growth. Relatively cautious fiscal and
foreign borrowing policies meant that serious debt crises were avoided,
Figure 10 PUBLICAND PRIVATEINVESTMENT
0.25
PrivateInvestment(as share of GDP)
0.2
0.15
0.1
i
0.05
tHPAS (w/oJapan)
-_ otherLMIEs
t
A
11
1974
1969
0.2
1979
1984
1989
PublicInvestment(as shareof GDP)
0.150.1-
0.05
ll
1969
1969
*HPA3s (w/o Japan)
-e-otheriLMIEs
l,,~~~~~~~~~~~~~~~~~~~~~~~~~~
1974
1979
Note: LMIEs = low- and middle-income
Source: World Bank data.
1984
economies.
1989
246 PAGE
which reduced the stop-go pattern of crisis and response that characterized many developing economies in the 1980s. Avoiding crisis and
the need for rescheduling meant that credit-worthiness was maintained, and it was easier to borrow in the short term and to avoid very
deep cuts, especially in investment.
In the 1970s overall levels of public investment did not differ markedly
between the HPAEs and other developing economies; over the decade
public investment rates in all economies rose from about 7% to 10%
(Figure 10). During the 1980s, public investment was higher in the
HPAEs than in other developing economies, but it had declined to
historical levels by the end of the decade. The HPAEs as a group
responded to the economic contraction of the 1980s by increasing public
investment above historically maintained levels. Sudden reductions in
aggregate demand and in investment compelled by debt crises were
major causes of the sharp declines in growth rates in many of the
heavily indebted economies of Latin America and elsewhere.
Better responses to macroeconomic shocks may also have resulted in
better measured productivity growth. A recent examination of the
sources of TFP change in developing countries concludes that in Latin
America and Africa, productivity levels increased relative to international best practice until 1973 and declined from 1973 onward, primarily
due to the inability of these economies to adapt to external shocks. The
decline was large enough in Latin America (and nearly large enough in
Africa) to offset gains due to technical progress (Friedberg, Khamis, and
Page, 1993).
The HPAEs, in contrast, are the only regional grouping of developing
economies that show steady improvement in productivity levels relative to international best practice. While macroeconomic contractions
are clearly visible in the HPAEs' pattern of movement of productivity
levels over time, the difference is the more rapid return to prior
productivity levels following the period of adjustment.
STRATEGIES
3.2 BROADLYBASEDEDUCATIONAL
In most of the economies of East Asia, public investments in education
were not only larger than elsewhere in absolute terms-they were also
better. They responded more appropriately to failures in the market for
education. Social rates of return to education are probably highest at
the primary level, where there are many positive externalities associated with literacy (Psacharopoulos, 1993). Capital market imperfections
and information problems, moreover, are also most severe at the primary level, reducing the scope for self-financed private systems. Returns from training at the university level, on the other hand, are
TheEastAsianMiracle?247
almost fully captured by the higher income of university graduates. All
of this argues for universal primary and broadly based secondary
education combined with restraint of public subsidies to higher education.
3.2.1 EducationalPerformancein the HPAEs This was precisely the educational strategy adopted by the HPAEs. Figures 11 and 12 present a
stylized summary of the results of regressing primary and secondary
enrollment rates on per capita national income for more than 90
developing economies for the years 1965 and 1987.25 Enrollment rates
are higher at higher levels of per capita income. But HPAE enrollment
rates have tended to be higher than predicted for their level of income.
At the primary level, this was most obvious in 1965, when Hong Kong,
Korea, Singapore, and Taiwan had already achieved universal primary
education, well ahead of other developing economies, and even Indonesia with its vast population had a primary enrollment rate above
70%.
By 1987, East Asia's superior education systems were evident at the
secondary level. Indonesia had a secondary enrollment rate of 46%,
well above other economies with roughly the same level of income, and
Korea had moved from 35% to 88%, maintaining its large lead in
relative performance. Only in Thailand was the 28% secondary enrollment rate well below the income-predicted 36% and the 54% mean for
middle-income economies.26 In recent years Thailand's weak educational performance has been felt, as serious shortages of educated
workers have begun to threaten continued very rapid growth. In part
as a function of their success in increasing enrollment, the East Asian
economies have also been faster to close the gap between male and
female enrollments.
A common, though imperfect measure of educational quality, is
expenditures per pupil. Between 1970 and 1989, real expenditures per
pupil at the primary level rose by 355% in Korea. In Mexico and Kenya,
expenditures rose by 64% and 38%, respectively, over the same time
period, and in Pakistan expenditures rose by only 13% between 1970
and 1985 (Birdsall and Sabot, 1993). These dramatic differences reflect
25. Behrman and Schneider (1992). The regressions control for a polynomial in average
per capita income in the relevant year. The authors used per capita GNP at official
exchange rates as the measure of income.
26. Despite increasing its secondary enrollment rate from 29% to 74%, Hong Kong also
fell well below the predicted level in 1987. This is because, at $15,000, the per capita
income of Hong Kong was so high that the OECD countries were now its comparators.
Figure 11 CROSS-ECONOMY REGRESSION FOR PRIMARYENROLLMENT RA
140
120 -
lindonesa (118)
137.01
Braz (10)
ial i
Singapore (105)
ip 21.91
Rep. ol Korea(1i
" 100
K
H
H15K
103)
)
129.2)f1i
3
Thaanad(95
1 P(
* -0
1.
Malaysa (90)
(179)
IC
E
/
80
e.0
H 1 --
60
' '
/
ngIadesh(59)
Pa
Thaian
.(40 d(7)
/
40 _
Pakistan
(52)
Bangladesh
(49)
1-14.9 1
1-14.91
iII/ *I Pakistan
I I (40)i II
1-24.5
',",'I,,'
20
0
*
A
1
'I,l ,'',tll
l l^^A I,lI,
I,l . . .
,,
vY
500
750
1,750
2,000
-
250
1-27.91
"I%#
.
iI
I
.I
I
?
i
i I i
500
PercapitaIncome(in1988U.S.dollars)
Note: Figures in parentheses are enrollment rates; bracketed numbers show residuals.
Source:Behrman and Schneider (1992).
I?
FORSECONDARY
REGRESSION
ENROLLMENT
Figure12 CROSS-ECONOMY
RA
E
c
J
o sla (46)
119.51
'P
0
0
t)
/
Pakslan(19)
Ia i-9.
/el
Bangladesh (18)
l-3.61
t
0
250
500
750
1,750
I
I
2,000
*
I
t
.
*
i
l
.
500
PercapitaIncome(in1988U.S.dollars)
Note: Figures in parentheses are enrollment rates; bracketed numbers show residuals.
Source: Behrman and Schneider (1992).
ll
250 PAGE
mostly differential changes over the period in income growth and in
the number of children entering schools, both of which favored the East
Asian economies. A somewhat better measure of school quality is the
performance of children on tests of cognitive skills, standardized across
economies. In the relatively few international comparisons available
from such tests, East Asian children tend to perform better than children from other developing regions-and
even, recently, better than
children from high income economies.27 Table 7 offers some comparative data on expenditures per pupil in the HPAEs and other countries
between 1965 and 1989.
3.2.2 Focusing Educational Spending Higher shares of national income
devoted to education cannot fully explain the larger accumulation of
human capital in the HPAEs. In both 1960 and 1989, public expenditure
on education as a percentage of GNP was not much higher in East Asia
than elsewhere (Table 8). In 1960 the share was 2.2% for all developing
economies, 2.4% for sub-Saharan Africa, and 2.5% for East Asia. Over
the three decades that followed, governments of East Asia markedly
increased the share of national output they invested in formal education. But so did governments in other developing regions. In 1989 the
share in Africa, 4.1%, was higher than the East Asian share, 3.7%, which
barely exceeded the average share for all developing economies, 3.6%.28
The allocation of public expenditure between basic and higher education is the major public policy factor that accounts for East Asia's
extraordinary performance with regard to the quantity of basic education provided. The share of public expenditure on education allocated
to basic education has been consistently higher in East Asia than
elsewhere. The share of public funds allocated to tertiary education in
East Asia has tended to be low, averaging roughly 15% over the
last three decades.29 In Latin America the share has been roughly
27. Birdsall and Sabot (1993) cite Stevenson and Stigler (1992), among others, who report
results of tests in cities of Japan, the United States, and Taiwan.
28. Government expenditure on education, expressed as a percentage of GNP, was used
as an explanatory variable in a cross-country regression in which expected years of
schooling of the school-age cohort (essentially an aggregate of enrollment rates) is the
independent variable. For a sample of 15 Asian and Latin American countries, the
expenditure variable was insignificant. See Tan and Mingat (1992).
29. In Korea and Taiwan the share of public expenditure on education allocated to higher
education has increased over the last decade or so for two reasons. On the one hand,
universal and near universal enrollment rates have been achieved at the primary and
secondary levels, respectively. On the other hand, the increase has been consistent
with the shift in the structure of production and exports to more technologically
sophisticated and skill-intensive products, and the consequent increase in the demand
for engineers and other skilled workers.
Table 7 PUBLIC EXPENDITURE PER STUDENT ON PRIMARYAND SECONDARY
1965
Economy
HPAEs
Hong Kong
Korea, Rep. of
Malaysia
Singapore
Thailand
Primary
198
1975
Secondary
Primary
Secondary
Primary
558.1
389.8
351.2
834.0
133.0
9.9
14.7
192.3
51.2
180.5
32.9
17.6
67.7
193.3
49.0
314.7
74.8
Other selectedeconomies
Brazil
Ghana
22.3
India
5.3
28.9
Kenya
Mexico
23.8
Pakistan
5.9
76.9
179.1
122.0
71.3
336.7
56.6
7.1
17.5
30.9
110.2
15.9
315.2
33.4
Note: Cells are empty where data are unavailable.
Source:Computed from UNESCO and World Bank data.
155.4
27.9
38.1
50.3
101.9
21.1
252 PAGE
Table 8
PUBLIC EXPENDITUREON EDUCATION
AS A PERCENTAGEOF GNP
1960
1989
HPAEs
Hong Kong
Korea, Rep. of
Singapore
Malaysia
Thailand
Indonesiaa
Averageb
2.0
2.8
2.9
2.3
2.5
2.5
2.8
3.6
3.4
5.6
3.2
0.9
3.7
Other
Brazil
Pakistan
Less developed
1.9
1.1
1.3
3.7
2.6
3.1
2.4
4.1
Economy/Region
economiesc
Sub-Saharan Africa
aAlternativesources of data indicate that expenditure on public
education as a percentage of GDP was 3.0%in Indonesia in 1989.
Average does not include Indonesia.
CLower-and middle-income economies.
Source:United Nations Development Program (1991).
24%.30In South Asia the share is close to the Latin American level. This
had been the case in Africa as well, but in recent years the share has
declined to East Asian levels.
The broad base of human capital was critically important to rapid
growth in the HPAEs. Because the HPAEs attained universal primary
education early, literacy was high and cognitive skill levels were substantially above those in other developing economies. Basic education
in the HPAEs is highly oriented to the acquisition of general academic
skills, while postsecondary education tends to be oriented toward
vocational skills. Some HPAEs have also been unusually large-scale
importers of educational services, particularly in vocationally and technologically sophisticated disciplines. Firms, therefore, had an easier
time upgrading the skills of their workers and mastering new technology. In addition, rapid human capital accumulation reduced income
inequality by increasing the relative abundance of educated workers,
thereby lowering the scarcity rents associated with cognitive skills.
30. Given the smaller size of the basic education age cohort in East Asia than in Latin
America, this difference underestimates the gap between the two regions in the
strength of the public sector's commitment to basic education.
The East Asian Miracle ?253
These benefits were particularly evident in the countryside (Birdsall
and Sabot, 1993).
4. UnconventionalWisdom-The ExportPush
and IndustrialPolicies
Most East Asian governments have pursued sector-specific industrial
policies to some degree. The best known instances include Japan's
heavy industry promotion policies of the 1950s and the subsequent
imitation of these policies in Korea. These policies included import
protection as well as subsidies for capital and other imported inputs.
Malaysia, Singapore, Taiwan, and even Hong Kong have also established programs-typically
with more moderate incentives-to
accelerate development of advanced industries. In the capital market,
governments intervened systematically both to control interest rates
and to direct credit.
Industrial targeting could have resulted in extensive rent seeking and
great inefficiency. Because it apparently did not, the success or failure of
selective interventions are among the most controversial aspects of the
East Asian success story. Some of the reasons why selective interventions do not appear to have had the disastrous results of those pursued
in other developing economies are straightforward. Labor market policies in the HPAEs tended to emphasize flexibility and competitive
determination of wages (Fields, 1993). Directed credit programs were
undertaken within a framework of generally low subsidies to borrowers
and careful monitoring (Vittas and Cho, 1993). As a consequence, the
relative prices of labor and capital in the HPAEs were closer to their
scarcity values than in other developing economies (World Bank, 1993).
Nevertheless, abundant evidence exists that, especially in the cases of
Japan, Korea, and Taiwan, governments made systematic efforts to alter
industrial structure for the purpose of accelerating productivity change.
The impact of these policies on the HPAEs growth remains a topic of
heated debate.31 This section proposes two policy lessons: first, that
promotion of manufactured exports was a significant source of measured TFP change, and second, that industrial policies mattered relatively little in the overall record of East Asia's extraordinary growth.
The upshot of these propositions is that export-not industrial-promotion was at the heart of the HPAEs' productivity performance.
31. See, e.g., the collection of essays in World Development (1994); Rodrik (1993),
critically
reviewing World Bank (1993).
254 PAGE
4.1 EXPORTGROWTHAND PRODUCTIVITY
CHANGE
Although all HPAEs except Hong Kong passed through an import
substitution phase, with high and variable protection of domestic import substitutes, these periods ended earlier than in other economies.
Hong Kong, Malaysia, and Singapore adopted trade regimes that were
close to free trade; Japan, Korea, and Taiwan adopted mixed regimes
that were largely free for export industries. Indonesia and Thailand
have in the 1980s begun to reduce protection. Exchange rate policies
were liberalized, and currencies frequently devalued, to support export
growth.32
Export push strategies were implemented in three very different
ways in the HPAEs. Hong Kong and Singapore established free-trade
regimes, linking their domestic prices to international prices; the export
push was an outcome of the very limited size of the domestic market
coupled with neutral incentives between producing for the domestic or
international market. Both economies made export credit available,
although they did not subsidize it, and Singapore focused its efforts on
attracting foreign investment in exporting firms.
In Japan, Korea, and Taiwan, incentives were essentially neutral on
average between import substitutes and exports. But within the tradedgoods sector, export incentives coexisted with substantial remaining
protection of the domestic market. Export incentives, moreover, were
not neutral among industries or firms. There was an effort in Japan,
Korea, Singapore, and Taiwan to promote specific exporting industries.
Protection was combined with either compulsion or strong incentives to
export. In Korea, firm specific export targets were employed; in Japan
and Taiwan access to subsidized export credit and undervaluation of
the currency acted as an offset to the protection of the local market.
What was important in all cases was that governments were credibly
committed to the promotion of manufactured exports. They adopted an
export push strategy.33
The HPAEs' export performance is reflected in their steadily rising
share of world exports (Table 9). As a group, the HPAEs increased their
share in world exports from 7.9% in 1965 to 13.1% in 1980 and 18.2% in
32. Corden (1993) refers to this as "exchange rate protection."
33. Colin Bradford was the first to introduce the concept of export push into the analysis
of East Asia's rapid growth. He has defined it as effective exchange rates for
exportables exceeding those for importables (Bradford, 1994). We use the term
somewhat more broadly to indicate a credible commitment to a policy regime that will
ultimately yield effective exchange rates for exportables equal to or greater than the
EER for importables.
TheEastAsianMiracle?255
EASTASIANECONOMIES,
SELECTED
Table9 EXPORTPENETRATION,
1965-1990
Sharein developing
economyexports
Sharein worldexports
Economy
Totalexports
Japan
Four tigersa
SoutheastAsian NIEsb
HPAESubtotal
All developing economies
World
Exportsof manufactures
Japan
Four tigersa
SoutheastAsian NIEsb
HPAESub-total
All developing economies
World
1965
1980
1990
5.0
7.0
9.0
1.5
1.5
7.9
24.2
3.8
2.2
13.1
28.7
6.7
2.4
18.2
19.8
100
100
100
7.8
11.6
11.8
1.5
0.1
9.4
11.1
5.3
0.4
17.3
11.8
7.9
1.5
21.3
12.9
100
100
100
1965
-
1980
1990
-
6.0
6.2
12.2
100.0
13.3
7.8
21.1
100.0
33.9
12.4
56.3
100.0
13.2
1.1
14.2
100
44.9
3.8
48.6
100
61.5
12.0
73.5
100
-
-
-
aRepublicof Korea, Hong Kong, Singapore, and Taiwan.
bIndonesia, Malaysia, and Thailand.
Source:UN Trade Systems data.
1990. Manufactured exports have provided most of this growth. From
1965 to 1990, Japan emerged as the world's biggest exporter of manufactured goods, increasing its share of the world market from nearly 8% to
almost 12%. In the 1970s and 1980s, the locus of growth shifted to the
four tigers, whose share of manufactured exports grew nearly four
times faster than Japan's (Table 9). Beginning around 1980, the three
southeast Asian HPAEs (Indonesia, Malaysia, and Thailand), historically
dependent on commodity exports, recorded a similar but so far smaller
surge in manufactured exports.
Some analysts have, with hindsight, attributed these achievements to
unique cultural and geographical circumstances. But there was little
evidence at the outset that East Asian economies would achieve such
spectacular results. In the 1950s even trade optimists were export
pessimists and did not anticipate that Korea's exports would grow four
times as fast as world trade over the next 30 years.34 One obvious effect
of rapid export growth has been a marked increase in the openness of
34. See, e.g., Little (1982).
256 PAGE
these economies as measured by the share of exports plus imports in
GDP (Table 10).
Most explanations of the link between TFP growth and exports
emphasize such static factors as improved resource allocation among
sectors (presumably arising from reductions in anti-export bias),
economies of scale and improved capacity utilization, which may account for a high level of productivity being achieved after a short
period of export orientation but not for continuing high TFP growth
rates. However, the relationship between exports and productivity
growth may arise from export's role in helping economies adopt and
master international best practice. Because firms that export have greater
access to best practice technology in imperfect world technology markets, there are both benefits to the enterprise and spillovers to the rest
of the economy that are not reflected in market transactions (Pack and
Page, 1993; Bradford, 1994). These information-related externalities can
be an important source of rapid TFP growth.
Trade and educational strategies, moreover, may have worked together. High levels of labor force cognitive skills permit better firm-level
adoption, adaptation, and mastery of technology (Birdsall and Sabot,
1993). It is doubtful that the HPAEs could have made as productive use
of foreign knowledge and imported capital and benefited as much from
embodiment without highly skilled domestic engineers and workers.
Table10 RATIOOF TOTALTRADETO GDP
Economy/ region
HPAEs
Hong Kong
Indonesia
Korea,Rep. of
Japan
Malaysia
Singapore
Taiwan
Thailand
Sub-SaharanAfrica
South Asia
LatinAmericaand
Caribbean
1970
1980
1985
1988
1.50
1.52
1.78
2.82
0.25
0.32
0.19
0.89
2.12
0.53
0.28
0.24
0.11
0.20
0.46
0.63
0.25
1.00
3.70
0.95
0.49
0.30
0.17
0.25
0.38
0.66
0.23
0.85
2.77
0.82
0.44
0.27
0.16
0.22
0.42
0.66
0.21
1.09
3.47
0.90
0.35
0.45
0.19
0.23
Note: Total trade = value of exports and value of imports/gross domestic product.
Sources: World Bank data; Taiwan, various issues; National Accounts Statistics: Analysis of Main
Aggregates, 1988-1989 (United Nations).
TheEastAsianMiracle?257
Thus, exports and human capital formation may have interacted to
provide a particularly rapid phase of productivity-based catch-up.
To test the hypothesis that export orientation played a significant role
in the HPAEs' TFP growth, we attempt to explain variations across
economies in TFP growth rates in terms of relative income, educational
attainment (as measured by the average stock of education per person),
a measure of trade orientation, and measures of manufactured export
performance (Table 11).
The relative level of GDP in 1960 is included in the explanatory
regression to test for the presence of conditional, productivity-based
catch-up. The interpretation of the variable is not straightforward,
however, since it also captures the resource allocation gains arising
from structural transformation at low levels of income. Because these
potential gains decline with increases in per capita income, the relative
income variable also captures the effect of structural change on the
one-sector TFP estimates (Pack, 1993).
Numerous efforts have been made to test the relationship between
"outward orientation" and productivity growth.35 Dollar (1990) uses
the international comparisons of price levels compiled for 121 market
countries by Summers and Heston (1988) to develop an index of
"outward orientation" for 95 developing countries. We employ Dollar's
index in our basic specification and find that there is a significant and
positive relationship between "outward orientation" and TFP growth.
The interpretation of this variable is also not straightforward, however. Rodrik (1993) has discussed the problems in the construction of
the index and its interpretation, and argues that it is best regarded as an
index of exchange rate mismanagement. To the extent that this is
correct, our results differ from those of Fischer (1993) in finding a
significant negative relationship between inappropriate exchange rate
policy and productivity growth.
Manufactured export performance is strongly correlated (at the 1%
level) with increased rates of TFP growth. We use two measures of
export orientation that have somewhat different interpretations. The
first is the share of manufactured exports in total exports. This is a
crude measure of the probability that the marginal export will be a
manufactured good. The second is the more conventional share of
manufactured exports in GDP, which measures manufactured export
concentration for the economy. Both the share of manufactured exports
in total exports and the share of manufactured exports in GDP are
35. See, e.g., Thomas and Wang (1993) and Nishimizu and Page (1991).
Table 11 DETERMINANTS OF TOTAL FACTOR PRODUCTIVITY GROWTH, 1960(DEPENDENT VARIABLE:RATE OF GROWTH OF TOTAL FACTOR PRO
Numberof observations
Intercept
GDP relative to U.S., 1960
Educational attainment, 1960
Dollar index
Average manufactured exports/
total exports, 1960-1985
Interaction term: Education attainment
1960 times manufactured exports
/total exports, 1960-1985
Average manufactured exports/GDP,
1965-1985
Interaction term: Education attainment
1960 times manufactured exports/
GDP, 1965-1985
Adjusted R2
56
- 83.1863**
(15.6575)
- 4.5638*
(2.1994)
0.3083**
(0.0951)
0.8328**
(0.1579)
0.4854
47
62
47
- 0.3584 -64.9123** -71.1186**
(0.2614)
(14.0585)
(14.8657)
- 3.3864** - 4.8047* - 5.5757*
(1.0189)
(1.9771)
(2.0637)
0.2158**
0.1471
0.0680
(0.0808)
(0.0874)
(0.1082)
0.6493**
0.7154*
(0.1417)
(0.1508)
0.0345**
0.0314**
0.0159
(0.0066)
(0.0058)
(0.0142)
0.0032
(0.0026)
0.4665
0.6333
*
Statistically significant at the 0.05 level.
**Statistically significant at the 0.01 level.
Note: Coefficient is top number. Standard error is bottom number in parentheses.
Source: World Bank staff estimates.
0.6376
TheEastAsianMiracle*259
significantly and positively correlated with TFP growth, controlling for
relative income and educational attainment.36
When the manufactured export variables are introduced together
with the Dollar index, both are significant, although the coefficient on
the share of manufactured exports in total output is less precisely
estimated. Our interpretation of these results is that controlling for
overall "outward orientation" (or appropriate exchange rate policy),
high manufactured export orientation increases TFP growth. A high
concentration of manufactured exports relative to total exports, rather
than the relative size of the manufactured export sector, is more closely
associated with productivity growth in a cross-economy framework.
This is consistent with the hypothesis that export-based learning is
more closely related to manufactured export orientation than to manufactured export volume.
The education stock variable is included as a crude test of educationally based externalities. We find, consistent with this hypothesis, that
the education stock variable is positively and significantly associated
with TFP growth when either the Dollar index of outward orientation
or either index of manufactured export orientation is used independently in the regressions. When both indices of outward orientation are
introduced jointly, however, the magnitude of the coefficient on educational attainment declines, as does its significance (to the .10 level). This
is suggestive of an interaction between trade orientation and educational attainment.
We also find some evidence of a positive interaction between the
share of manufactured exports in total exports and in national income
and the stock of education. The coefficient of the interaction term
between these two variables is positive but not significant at conventional levels, and the export share variable becomes insignificant. When
we consider the contribution of the variables taken together to explaining the variation in TFP growth rates, however, it is statistically positive. We conclude that export performance and education interact
36. The evidence from our cross-economyestimatesis supportedby a number of recent
microeconomicstudies that attemptto test the link between exportsand productivity
growth. Pack and Page (1993) present evidence from Koreaand Taiwan that at the
sectoral level, rapid export growth is correlated with the pattern of productivity
change; exporting sectors have higher sectoralrates of TFPgrowth. Wei (1993) uses
city level data from China and finds a statisticallysignificantrelationshipbetween
export growth and productivity growth. Aw and Hwang (1993), using firm-level
microeconomicdata from Taiwan,find a statisticallysignificantrelationshipbetween
productivitylevel differencesamong manufacturingfirmsand export orientation.
260 PAGE
positively; higher levels of education raise the contribution of manufactured export concentration to TFP growth.37
OF INDUSTRIALPOLICY
4.2 THEINSIGNIFICANCE
Proponents of neutrality and intervention both cite the high performing East Asian economies as evidence supporting their views. Balassa
(1991), Krueger (1993), Hughes (1992), and others argue that openness
to international trade, based on largely neutral incentives, was the
critical factor in East Asia's rapid growth. On the other hand, advocates
of interventions, while acknowledging the importance of trade, note
that incentives deriving from quantitative restrictions on imports, tariffs,
and subsidies were not neutral among sectors (or firms) during the
HPAEs' periods of rapid growth. They argue that the HPAE governments successfully intervened to change comparative advantage
(Amsden, 1989; Wade, 1990; Singh, 1992).
Industrial policy interventions, which often use trade policy instruments, are motivated by the belief that shifting industrial structures
toward newer and more modern sectors increases the opportunities for
capturing the dynamic scale economies that result from learning. We
define industrial policies, as distinct from trade policies, as government
efforts to alter industrial structure to promote productivity-based
growth. This productivity-based growth may derive from learning,
technological innovation, or catch-up to international best practice.
All of the HPAEs, except Hong Kong, have employed industrial
policies as defined earlier. Japan and Korea had the most systematic set
of policies to alter industrial structure. Taiwan's efforts were less systematic but were nonetheless widespread. Industrial policy in Singapore was more functionally directed at the rapid upgrading of technology by direct foreign investors, regardless of type of output. Malaysia,
Indonesia, and Thailand have all used industrial policies but much less
systematically than the northeastern HPAEs.
37. We ran two joint F tests on the regressionsincluding the interactiveterm.The F tests
were both consistent with a high degree of multicollinearitybetween the interaction
term and the export term. The F Test rejects the null hypothesis that, taken jointly,
the coefficientson the three variables(education,manufacturedexports/total exports,
and the interactiveterm) are not significantlydifferentfrom zero (taken together, the
three are significantlydifferentfrom zero at the .01 level). Likewise,the F test rejects
the hypothesis that, taken jointly, the interactiveterm and the exportvariableare not
significantlydifferent from zero (taken together, the two are significantlydifferent
from zero at the .01 level). Where there is a high degree of multicollinearitybetween
the variables,the coefficientson the interactionterm and the export variable,despite
being separatelyinsignificant,should still be treated as best point estimates.
TheEastAsianMiracle?261
COMPOSITIONON MANUFACTURING
Table12 EFFECTOF SECTORAL
WIDEGROWTHOF TFP
Economy
Korea,Rep. of, 1966-1985
Japan,1960-1979
TFPgrowth,
actualvalueaddedweights
TFPgrowth,
adjustedweights
6.7
2.3
6.1
1.9
Note: Weighted by value added shares that would have prevailed if the metal products and
machinery sector had conformed to that predicted by the equations estimated by Syrquin and
Chenery (1989).
Sources:Pack (1993). Based on Kuroda, Korgenson, and Nizhimizu (1985) for TFP estimates for
Japan.
4.2.1 Industrial Growth and Productivity Change HPAE industrial growth
patterns differ from the patterns in most other low- and middle-income
economies in the relative size and growth rates of two important
industrial subsectors-metal products, electronics and machinery, and
textiles and garments. Table 12 shows the ratio of the share of the value
added in five key International Standard Industrial Classification (ISIC)
subsectors as a percentage of the total value added of manufacturing to
cross-country norms (Chenery, 1987). Among the HPAEs, metal products, electronics, and machinery (ISIC subsector 38, or MPM) have
grown unusually fast. The sector's share of manufacturing value added
doubled in Singapore and Japan, nearly tripled in Korea, and Indonesia, and quadrupled in Malaysia. More surprising than the importance
of growth in MPM, which provides vital inputs to numerous other
manufacturing subsectors, is continued importance of textiles and garments even as the rapidly developing Asian economies shifted from
labor intensive to capital intensive production.
Detailed sectoral growth rates of TFP are available for Japan, Korea,
and Taiwan. There is both good news and bad news for advocates of
industrial policy in the productivity performance of East Asian industry.
The good news is that, on average, rates of productivity change in
industry in Japan (before 1973), Korea, and Taiwan, which are the only
economies for which we have detailed sectoral estimates of TFP growth,
were high by international standards (Page, 1990).38 There are now
sufficiently long time series data to conclude that in these economies,
TFP growth has accounted for a substantial fraction of the growth of
38. Young (1993) disputes even this assertion, although his sample of LDC comparators is
small.
262 PAGE
constant price value added in manufacturing. Given the length of time
of the observations, it seems unlikely that the measured growth rates of
TFP could be attributable to cyclical phenomena or growing capacity
utilization of initial large investments.
The bad news is that, in general, productivity change has not been
higher in promoted sectors. Japan may be an exception. Between 1960
and 1979, chemicals and the metal working machinery complex had
unusually good TFP performance (Jorgenson, Kuroda, and Nishimizu,
1987). Japan's industrial structure differs from international norms in
these sectors and exhibits quite high values of the share of value added
in total manufacturing. These industries are those that observers usually
point to as having received significant government support including
efforts to stimulate productivity growth.
A number of calculations of TFP have been carried out for Korea for a
variety of periods (Dollar and Sokoloff, 1990; Lim, 1991). From these
studies a number of patterns can be identified that are broadly consistent with one another. What is most striking are the high values of TFP
change in most sectors by international standards (Nishimizu and Page,
1991). Although the Korean government selectively promoted chemicals
and iron and steel (included in basic metals), the large growth in the
share of iron and steel was accompanied by quite low TFP performance
between 1966 and 1985; textiles and clothing, on the other hand, had
very high rates of TFP growth. The promoted chemical sector, whose
relative size was decreasing, was characterized by considerably higher
than average TFP growth during this same period.
The government in Taiwan did not attempt to influence sectoral
evolution as strongly as the government of Korea. Nevertheless, there
was substantial effort devoted to encouraging specific sectors, particularly those viewed as either capital- or technology-intensive. However,
there is no relationship between capital intensity and productivity
change at the sectoral level. In fact, the highest sectoral rates of TFP
change are recorded in textiles and apparel.
Overall, the evidence that industrial policy systematically promoted
sectors with high productivity change is weak. In Japan there is some
support for the assertion that TFP growth was higher in selected
sectors, while in Korea and Taiwan, activities that were not promoted
(e.g., textiles) had equally as impressive TFP performances as those that
were. The critical empirical question, of course, is whether, given any
plausible assumptions regarding the relative size of promoted sectors in
the absence of industrial policy and their observed rates of productivity
change, industrial policies accelerate the overall rate of productivity
change in manufacturing.
TheEastAsianMiracle?263
4.2.2 The Effect of Industrial Policy on Manufacturing Productivity Growth
Aggregate TFP in any period can be decomposed by weighing each
sector's level of total factor productivity, A, t, by the sector's share in
value added, vi,t. The growth of TFP will then depend on changes in
A, t and changes in vi, . Algebraically, this relation can be written as
A Log A =
(vit log Ai, t -
i, t-_ log Ai, t-1).
(3)
Equation (3) gives the growth in A due to the change in productivity of
sectors assuming constant sectoral shares and/or the growth in the
value added share of sectors, whose productivity is high.
Of the countries considered here, Japan and Korea are the countries
in which activist industrial policies were employed most consistently to
achieve productivity-based catch-up. On the basis of international comparisons, two sectors were generally overrepresented in the industrial
sectors of the HPAEs, metal products and machinery, and textiles and
apparel. Broadly speaking, only the former sector was promoted in
HPAEs that used industrial policy. One test of the significance of
industrial policy, then, is whether, given the observed values of sectoral
TFP growth, the unusual evolution of the sectoral pattern of production
in some of the countries had a significant quantitative impact on overall
TFP growth.39
To answer this we first calculate the manufacturing-wide growth of
A* using the observed value-added shares at the end of the period of
1966 to 1985, and the A* values estimated for Korea by Lim (1991) and
for Japan by Jorgenson, Kuroda, and Nishimizu (1987). We then recalculate the value added weights assuming that the sectors in ISIC 38 had
been at their predicted value, on the basis of international norms taken
from Chenery (1986), and we reassign the residual value added to all
other sectors equally. Thus, in Korea, instead of accounting for 38% of
manufacturing value added, sector 38 accounts for 14%.40
The result for Korea is shown in row 1 in Table 12. The actual sectorwide growth rate, A*, was 6.7%, the recalculated one 6.1%, during a
39. It is possible that industrial policy significantly increased the values of A* in sectors
other than those promoted due to one or more externalities. While this may be the
case, it is likely that most externalities occur within individual sectors or in closely
related ones. Given the many branches constituting the Metal Products and Machinery sector, the externalities should have been revealed in its own TFP value.
40. Put differently, it has been shown earlier that Korea's vP/vA ratio for MPP was 2.76.
If this had been one, the actual share of value added would have been 13% rather
than 36%, i.e., the share would have been constant at the 1968 level. If factors had
been allocated to all of the remaining sectors equally, the sector-wide average value
for A* would have been .061 rather than .067.
264 PAGE
period in which the growth rate of manufacturing value added was
17.5% per annum. Even if the "excess" growth of the MPM sector is
attributed to selective intervention, its rate of TFP growth was not
sufficiently above that of other sectors to make a large contribution to
overall industrial TFP growth. The major reason for Korea's manufacturing success lay in high individual TFP growth rates for most sectors
in most periods. A calculation, similar to that for Korea can be done for
Japan. Table 12 shows the actual growth of A* for 1960-1979 and the
estimated growth, had the sectors constituting 38 been at the international norm, 24% of manufacturing value added rather than 41% in
1979. The value of A* would have declined from 2.3 to 1.9, a relatively
small decrease given the growth rate of manufacturing value added of
8.7% per annum.
5. Policiesfor RapidGrowthin a Changing
WorldEconomy
What caused East Asia's success? In large measure the HPAEs achieved
high growth by getting the basics right. Private domestic investment,
combined with rapidly growing human capital, were the principal
engines of growth. Agriculture, while declining in relative importance,
experienced rapid growth and productivity change. Population growth
rates declined more rapidly in the HPAEs than in other parts of the
developing world. And some of these economies also benefited from a
head start in terms of the education of the labor force and capable and
effective systems of public administration. In this sense there is little
that is "miraculous" about the HPAEs' superior record of growth; it is
largely due to superior accumulation.
To what extent are those lessons from East Asia applicable to other
developing economies? In the HPAEs, a wide variety of policies, across
countries and over time within countries, were used to achieve the
critical functions of growth: accumulation, allocation, and productivity
growth. While the sheer variety of policies precludes drawing any
simple lessons or making any simple recommendations, two fundamentalist lessons can be drawn. Macroeconomic stability and the capacity to
respond effectively to macroeconomic shocks helped to accelerate
growth through all three mechanisms, increasing accumulation, improving resource allocation, and increasing productivity growth. Education policies that stressed broadly based primary and secondary
education contributed directly to output growth and apparently also
indirectly through the interaction of educational shocks and export
orientation to TFP growth.
TheEastAsianMiracle?265
Our judgment was that the promotion of specific individual industries made relatively little difference to the HPAEs' success. Rather,
export push strategies have been the most generally successful selective
approach used by the HPAEs and hold the greatest promise for other
developing economies. It is not altogether surprising that we conclude
that export orientation rather than industrial policy was mainly responsible for improving productivity growth in Japan and Korea. These
economies, although selectively promoting capital and knowledge intensive industries, still aimed at creating profitable, internationally competitive firms. The yardstick used to evaluate industrial policies success
-mainly export performance-provided a market test of the success or
failure of the policy instruments chosen. Picking winners may have
succeeded more because Japan and Korea set export targets for promoted industries and used export performance to assess the success of
policies than because of success in selecting industrial subsectors.
Despite their potential benefits, selective interventions, especially
when combined with access to scarce resources such as foreign exchange and credit, have a high risk of capture by the participants.
resource allocation may have been imCapture was avoided-and
and
Korea
proved-in Japan
by the combination of repeated relationbetween
business
and
ships
government and the use of exports, not
domestic GDP growth, as the yardstick by which the success of the
industrial strategy was measured. In other HPAEs, however, such
elaborate contests were not used-e.g., in Hong Kong, Singapore, or
Taiwan-or were unsuccessful, as in the cases of the HICOM drive in
Malaysia or recent efforts at technological leapfrogging in Indonesia.
In the HPAEs that intervened selectively to promote exports, a
contest based on performance in global markets played the allocative
role that is normally ascribed to neutral exposure of both import
substituting and exporting industries to international competition. But
these contest-based incentive structures required high government institutional capability. One of the keys to success of the export push in
some of the HPAEs, especially Japan and Korea, was the government's
ability to combine cooperation with competition. They were able to do
this first, because their civil services and public institutions were largely
staffed by competent and honest civil servants, and second, because
firms and bureaucrats knew that there was a single yardstick for
performance, exports.
Export targets provided a consistent yardstick to measure the success
of market interventions. When protected sectors interfered with the
exports of other sectors, the latter could seek redress and were successful. Even where domestic content rules were imposed, e.g., on foreign
266 PAGE
direct investors in Taiwan, they were suspended if they interfered with
exports. The emphasis on export competitiveness gave businesses and
bureaucrats a transparent and objective system to gauge the desirability
of specific actions. Interventions could not be made arbitrarily, because
these could be appealed at a higher level of government if they
interfered with exports. The more recent export push efforts of the
Southeast Asian NIEs have relied less on highly specific incentives and
more on gradual reductions in import protection, coupled with institutional support of exporters and a duty-free regime for inputs into
exports. These "GATT friendly" export promotion strategies offer substantial scope for adoption by other developing economies.
REFERENCES
New
Amsden, A. H. (1989). Asia's nextgiant:SouthKoreaandlateindustrialization.
York:Oxford University Press.
(1993). Why isn't the whole world experimenting with the East Asian
model to develop? Review of the World Bank's EastAsian MiracleReportto
appear in "Symposium on the World Bank's East Asian Miracle Report," in
WorldDevelopment.
Aw, B.-Y., and A. R. Hwang. (1993). Productivity and the export market: A firm
level analysis. Pennsylvania State University, State College, PA.
Balassa, B. (1988). The lessons of East Asian development. EconomicDevelopment
and Cultural Change 36: (supplement):S273-S290.
(1991). Economicpoliciesin the PacificAreadevelopingcountries.New York:
New YorkUniversity Press.
Barro,R. J. (1989). A cross-countrystudy of growth, savings, and government.
NBERworking paper No. 2855, Cambridge,MA.
, and J.-W.Lee. (1993). Internationalcomparisonsof educational attainment. Paper presented at the conference, "How do national policies affect
long-run growth?" Washington, DC: The World Bank. February.
Behrman, J., and R. Schneider. (1992). An international perspective on schooling
investment in the last quarter century in some fast-growing eastern and
southeastern countries. Background paper for The East Asian Miracle. Washington, DC: Policy Research Department, World Bank.
Birdsall, N., and R. H. Sabot. (1993). Virtuous circles: Human capital growth and
equity in East Asia. Background paper for The East Asian Miracle, Washington,
DC: World Bank, Policy ResearchDepartment.
Bradford,C. I. (1994).From trade-drivengrowth to growth-driventrade: Reappraising the East Asian development experience.Washington,DC: Organisation for Economic Co-Operation and Development.
and
Chenery, H. B. (1986). "Growth and transformation."In Industrialization
growth: A comparativestudy. H. Chenery, S. Robinson, and M. Syrquin (eds.).
London: Oxford University Press.
Collins, S. and W. Park. (1989). External debt and macroeconomic performance
vol. 3.
in South Korea. In Developingcountrydebtand economicperformance,
J. Sachs and S. Collins (eds.). Chicago: University of Chicago Press.
The East Asian Miracle *267
Corden, W. M. (1993). Seven Asian miracle economies: Overview of macroeconomic policies. Background paper for The East Asian Miracle. Washington,
DC: Policy Research Department, World Bank.
Dollar, D. (1990). Outward orientation and growth: An empirical study using a
price-based measure of openness. Washington, DC: The World Bank.
(1991). Exploiting the advantages of backwardness: The importance of
education and outward orientation. Washington, DC: The World Bank.
(1992). Outward-oriented developing economies really do grow more
rapidly: Evidence from 95 LDCs, 1976-1985. Economic Development and Cultural Change 40:523-544.
, and K. Sokoloff. (1990). Patterns of productivity growth in South
Korean manufacturing industries, 1963-79. Journal of Development Economics
33:309-327.
Dorbusch, R., and S. Fischer. (1993). Moderate inflation. World Bank Economic
Review 7:1-44.
Easterly, W., M. Kremer, L. Pritchett, and L. Summers. (1993). Good policy or
good luck? Country growth performance and temporary shocks. Paper presented at the conference, "How do national policies affect long-run growth?"
Washington, DC: The World Bank. February.
, C. Rodriguez, and K. Schmidt-Hebbel (eds.). Public sector deficits and
macroeconomicperformance.New York: Oxford University Press, forthcoming.
Edwards, S. (1992). Trade orientation, distortions and growth in developing
countries. Journal of DevelopmentEconomics39:31-57.
Elias, V. (1991). The role of total factor productivity on economic growth.
Background paper for the World Development Report, 1991. Washington, DC:
Office of the Vice President, Development Economics, The World Bank.
Fields, G. (1993). Changing labor market conditions and economic development
in Hong Kong, Korea, Singapore, and Taiwan. Background paper for The East
Asian Miracle. Washington, DC: Policy Research Department, The World
Bank.
Fischer, S. (1993). "Macroeconomic Factors in Growth." Paper presented at the
conference, "How Do National Policies Affect Long-Run Growth?" Washington, DC: The World Bank, February.
Friedberg, L., M. Khamis, and J. M. Page, Jr. (1993). Productivity change and
technical efficiency in the high performing Asian economies. Washington,
DC: The World Bank.
Hughes, Helen. (1992). "East Asian Export Success." Canberra: Research School
of Pacific Studies, National Australian University.
IMF (International Monetary Fund). (1993). World Economic Outlook. Washington, DC, May.
Jorgenson, D. W., M. Kuroda, and M. Nishimizu. (1987). Japan-U.S. industry-level
productivity comparisons, 1960-70. Journal of the Japanese and International
Economics 1:1-30.
Krueger, A. 0. (1990). Perspectiveson trade and development.Chicago: University
of Chicago Press.
(1993). East Asia: Lessons for growth theory. Paper presented at the 4th
Annual East Asian Seminar on Economics, San Francisco, CA, June 17-19,
1993, National Bureau of Economic Research.
268- PAGE
Lim, Y. (1991). Upgrading industrial base and trade: The role of government
policy in Korea. New York: United Nations Industrial Development Organization.
Little, I. (1982). Economicdevelopment.New York: Basic Books, Inc.
Nehru, V., and A. Dhareshwar (1993). A new database on physical capital stock:
sources, methodology, and results. Washington, DC: The World Bank.
Nishimizu, M., and J. M. Page, Jr. (1987). Economic policies and productivity
change in industry: An international comparison. Paper for World Bank
Research Project, Productivity Change in Infant Industry. Washington, DC:
World Bank.
(1991). "Trade policy, market orientation and productivity
, and
change in industry." In Trade theory and economicreform:North, south and east
-essays in honor of Bela Balassa. Jaime de Melo and Andres Sapir (eds.).
Oxford: Basil Blackwell.
Pack, H. (1993). Industrial and trade policies in the high performing Asian
economies. Background paper for The East Asian Miracle. Washington, DC:
Policy Research Department, The World Bank.
, and J. M. Page, Jr. (1993). Accumulation, exports and growth in the high
performing Asian economies. Background paper for The East Asian Miracle,
presented at the Carnegie-Rochester Conference on Public Policy.
Page, J. M., Jr. (1990). The pursuit of industrial growth: Policy initiatives and
economic consequences. In Public policy and economic development:Essays in
honor of Ian Little. M. F. G. Scott and D. Lal (eds.). Oxford: Clarendon Press.
Pfeffermann, G. P., and A. Madarassy. (1992). Trends in private investment in
developing countries, 1992 ed. IFC discussion paper 14. Washington, DC: The
World Bank and IFC.
Psacharopoulos, G. (1993). Returns to investment in education: A global update.
Policy Research working paper no. 1067, Latin America and the Caribbean
Region. Washington, DC: The World Bank.
Rodrik, D. (1993). King Kong meets Godzilla: The World Bank and The East
Asian miracle. Paper prepared for the Overseas Development Council, New
York.
Romer, P. M. (1987). Crazy explanations for the productivity slowdown. In
NBER MacroeconomicsAnnual, Cambridge, MA: The MIT Press.
Singh, A. (1992). "Close" vs. "strategic" integration with the world economy
and the "market-friendly approach to development" vs. an "industrial
policy": A critique of the World Development Report 1991 and an alternative
policy perspective. Cambridge, England: Faculty of Economics, University of
Cambridge.
Stevenson, H., and J. Stigler. (1992). The learning gap. New York: Summit Books.
Summers, R., and A. Heston. (1988). A new set of international comparisons of
real product and prices: Estimates for 130 countries. Review of Income and
Wealth 34:1-25.
Syrquin, M., and H. B. Chenery. (1989). Patterns of development, 1950 to 1983.
World Bank discussion paper 41. Washington, DC: The World Bank.
Taiwan, China. (1992). Statistical Yearbookof the Republicof China. Taipei: Directorate-General of Budget, Accounting, and Statistics.
Tan, J.-P., and A. Mingat. (1992). Educationin Asia: A comparativestudy of cost and
financing. Washington, DC: The World Bank.
Comment 269
Thomas,V., and Y. Wang. (1993).Governmentpolicies and productivitygrowth:
Is East Asia an exception? Background paper for The East Asian Miracle.
Washington,DC: Policy ResearchDepartment,The World Bank.
United Nations Development Programme.(1991). Human developmentreport.
New York:Oxford University Press.
Vittas, D., and Y. Je Cho. (1993). Credit policies: Lessons from East Asia.
Washington,DC: The World Bank.
Wade, R. (1990). Governingthe market:Economictheoryand the roleof the governmentin EastAsian industrialization.
Princeton,NJ: PrincetonUniversity Press.
Warr,P., and B. Nadhiprabha.(1993).Macroeconomicpolicies, crisisand growth
in the long-run: Thailand. Background paper prepared for Boom, Crisis,
Adjustment:TheMacroeconomic
Experienceof DevelopingCountries.New York:
Oxford University Press.
Wei, S.-Jin.(1993).Open door policy and China's rapid growth: Evidence from
city-level data. Prepared for the 4th Annual East Asian Seminar on Economics, the National Bureau of EconomicResearch,Cambridge,MA.
World Bank. (1992). Worlddevelopmentreport.New York: Oxford University
Press.
(1993). TheEastAsianmiracle.Washington,DC: Policy ResearchDepartment, The World Bank.
WorldDevelopment22: 627-670. (1994).
Young, A. (1992).A tale of two cities:Factoraccumulationand technicalchange
in Hong Kong and Singapore. Paper presented at a World Bank seminar.
February.
(1993). Lessons from the East Asian NICS:A contrarianview. Working
paper series no. 4482. Cambridge,MA: National Bureau of Economic Research, Inc.
Comment
SUSANM. COLLINS
Georgetown University, The Brookings Institution, and NBER
This provocative paper reexamines the experiences of the East Asian
miracle economies (so-called because of their remarkable growth performances). One relatively noncontroversial section of the paper provides
further evidence in support of two conventional conclusions-that
sound macroeconomic management as well as broad-based educational
policies played important roles in the impressive performances. The
provocative part of the paper focuses on the role of micro-management,
concluding that while export promotion did contribute to high growth,
the targeted industrial policies many of the countries employed did not.
270 ?COLLINS
Unfortunately, I am not convinced that the methodology used here
supports the author's conclusions. In my view, the jury is still out as to
the lessons East Asian economies teach about the merits of micro
interventions.
Before discussing the specifics of the paper, let me make two points.
First,the paper addresses very importantissues that it is helpful to put
in context. There is a long-standing debate about the role that government intervention has played in high-performing Asian economies
(HPAEs).This study comes out of a majorWorld Bank project to assess
the development strategy in East Asia and draw lessons for other
countries with poor growth histories. In particular,was there a reason
to amend the traditional World Bank view that such intervention is
harmful? Thus, the present paper and the larger World Bank study
seem to represent a qualitative change in World Bank doctrine, which
now allows a limited role for micro-managementas part of a sensible
development strategy. In my view, this more balanced "conventional
wisdom" is warranted and appropriate.
Second, based on my own knowledge of the Koreanexperience, I am
quite sympathetic to the view that the extensive interventions have
helped to foster growth in East Asia-in fact, I believe that this view is
probably correct. My critique of this paper should be taken as an
assessment of the specific approach and conclusions presented here,
and not as an outgrowth of my priors about the role of intervention.
Let me now return to the specifics of the paper, roughly in the order
of presentation. The first part of the paper is devoted to documenting
the experience of rapid growth among HPAEsand to examining alternative explanationsfor it. The author divides the views into two camps:
fundamentalists-who stress the role of factor accumulation and tend
to believe that markets worked well-versus mystics-who focus on
the role of technology and tend to believe that activist government
policies helped to remedy market failures.The author's analysis of the
data leads him to conclude that "the East Asian success story is, then,
primarilya fundamentalistone."
The analysis presented does not really allow one to distinguish
between the two views. It seems to me that the approach is likely to
underestimate the role of technology, making it difficult to draw firm
conclusions-if the objective is to debunk the mystic view, surely one
would like to know the maximum amount that technology could
reasonably contribute to growth. The paper estimates total factor productivity (TFP)for HPAEsunder the assumption that they all have the
same production function as industrial countries. It seems more reason-
Comment ?271
able to assume that their production functions were converging toward
the function for industrial countries. In fact, the paper tells us that the
elasticity of output with respect to capital is much larger for the
industrial country sample than for the full sample including the developing countries. This suggests that the methodology overstates the
contribution of capital accumulation to growth in the developing country HPAEs, thus understating the contribution of TFP. However, it is
difficult to tell how large the underestimate might be.
It is also difficult to draw strong conclusions from the evidence about
factor accumulation here. To study the role of accumulation, the paper
presents scatter plots of average income growth, investment rates, and
education indicators during 1960-1985 each relative to 1960 per capita
incomes for HPAEs and a group of comparator countries. Although the
HPAEs are outliers in terms of their growth rates, it is much less
obvious that they are outliers in terms of investment and education.
One problem is that these very aggregate measures of factor accumulation may mask important differences among countries. Further, the
method seems quite sensitive to time period. For example, Korea had
very low investment rates (12.1% of GDP) during 1960-1965 versus
27.2% during 1965-1985. If the period selected began in 1965, Korea
might be an "outlier" in terms of both growth and investment rates. I
also note that Denison-style growth decompositions do point to the
importance of factor accumulation in Korea. Kim and Park (1985)1 find
that increases in factor inputs contributed substantially to growth in
Korea during 1963-1982 (4.9% per annum).
It seems to me that both accumulation and technological progress
have been important-as a group, the HPAEs stand out because they
have tended to do well in both. (Of course, there are differences across
countries.) Perhaps the fundamentalist-mystic debate creates a false
dichotomy. Technical progress and accumulation appear to interact in
complex but difficult to measure ways, and we may learn a considerable
amount by studying this interaction.
The next section of the paper reaffirms two conventional lessons,
both of which I agree with strongly. I mention these lessons only
briefly. First, sound macroeconomic management produced a stable
environment for economic activity to flourish. In fact, over the past
decade, there has been a striking convergence of views about the
1. See Kim,K. S., and J. K. Park.(1985).Sourcesof economic
growthin Korea:1963-82. Seoul,
Korea: Korea Development Institute, pp. 67-69.
272 ?COLLINS
importance of small budget deficits, monetary control, and realistic
exchange rates. The experiences of these high-growth economies have
played a role in the development of this consensus. At the same time,
the paper does not address the political economy of how and why
these economies were able to initiate, implement, and follow through
on sound macroeconomic policies.2
Second, the paper emphasizes the role of broad-based education
systems. Here, I wonder if the point could have been made even more
strongly if data on private spending on education had been included
along with the public expenditure figures. In many Asian economies a
strikingly large fraction of disposable household income is devoted to
tutors and other additional education expenditures for children. In
addition, the paper notes that the share of national income devoted to
education during both 1960 and 1989 in HPAEs is similar to the share in
comparator countries. This is a little misleading, because the HPAEs
have had a dramatic decline in dependency rates between 1960 and
1989-hence, the rise in per pupil expenditures.
The last part of the paper draws two "unconventional" conclusions:
that the promotion of manufactured exports was a significant source of
measured TFP change, while industrial policies mattered little. In the
remainder of my remarks, I explain why the jury is still out on the
effects of these types of policies.
To study export promotion, the paper presents a cross-country regression of average TFP growth on the ratio of manufactured to total
exports, a measure of "openness" and other right-hand-side variables.
The export ratio and openness are found to be significant. However,
what does this tell us about the effects of active interventionism that
promoted certain types of exports through direct credit allocations and
other forms of micro management? The export ratio is partially a
structural variable that depends on endowments as well as on policy.
Further, as the paper notes, the measure of openness used is typically
described as a measure of sensible exchange rate management. Perhaps
this variable is simply picking up the importance of sound macroeconomic policy-a "conventional" lesson.
To study industrial policy, the paper compares TFP growth across
two-digit industries within countries. It finds that although the average
2. For example, see Haggard, S. and R. Kaufman(eds.). (1992). The politicsof economic
adjustment.Princeton,NJ: Princeton University Press; and Williamson,J. (ed.). (1994).
Thepoliticaleconomyof policyreform.Washington,DC: Institute for InternationalEconomics.
Comment 273
TFP is high among HPAEs, it does not appear to be relatively higher in
certain promoted sectors than in nonpromoted ones. However, it is not
clear that the promoted/nonpromoted
distinction is made appropriately. For instance, textiles and garments are taken to be a sector that
was not promoted, even though Korea promoted it during part of the
period under consideration. Further, perhaps a successful industrial
policy works by raising average sectoral productivity and reducing the
variation in productivity growth across sectors. This finding may be
quite consistent with an effective intervention strategy.
There are three additional problems with this analysis-as well as the
concern expressed earlier about how TFP is measured. First, the attempted distinction between export promotion and industrial policy is
strange. Industrial policies are defined as "distinct from trade policies,
government efforts to alter industrial structure to promote productivity
based growth." How would the author classify selective credit allocations to firms producing color televisions that helped to increase their
productivity and their exports? Surely such an intervention should
count as both industrial policy and export promotion!
Second, there is the problem of the "appropriate counterfactual." It is
not at all clear that other countries (with a variety of different characteristics) are the appropriate benchmark against which to measure the
implications of a given country's attempt to promote a particular group
of exports. Also, as already discussed, "good" industrial policy may
work to reduce differences in cross-sectoral productivity instead of
increasing them. These issues warranted considerably more discussion
in the paper. A clearer analytic framework that spelled out the channels
through which export promotion and industrial policies are expected to
work would have helped in this regard.
Finally, the methodology tries to look for broad generalizations,
when surely the HPAE and other experiences suggest that some types
of interventions work under some conditions. Thus, it would be instructive to focus our research on trying to understand better when particular types of interventions are most likely to work well, and when
poorly. For example, does sector matter-is it easier to succeed in
promoting certain types of industries? When are direct credit allocations riskier than tax-based incentives?
Overall, this paper has reinforced my view that sound macroeconomic policies and educational emphasis promote growth. However, it
has not altered my view that while the upside potential from an activist
intervention strategy appears to be high, the down-side risk appears to
be even more considerable.
274 *ITO
Comment
TAKATOSHI ITO
Harvard University and NBER
This paper presents Dr. Page's view on how rapid economic growth
occurred in the eight East Asian countries-Japan, Hong Kong, Korea,
Singapore, Taiwan, Indonesia, Malaysia, and Thailand. His knowledge
and analyses are mainly based on a grand research study conducted at
the World Bank (1993).1 The study at the World Bank is considered to
be an overture to second thoughts on its development assistant strategy. The value and virtue of government intervention, such as policy
loans (subsidized loan to targeted industries), is at the heart of the
controversy. This paper succinctly summarizes major debate points in
the literature of East Asian growth experience. My comments follow the
presentation of Page's paper.
1. RapidGrowthand Its Source
No one disagrees that growth experiences in the eight East Asian
countries were miraculous. All of the eight countries have recorded
growth in GDP per capita exceeding 4% a year, especially the "Four
Tigers" exceeding 6% a year from 1965 to 1989. This presents "outliers"
in the so-called convergence regression. Figure 1 of the paper clearly
shows that these East Asian countries are outliers in the regression of
growth on the initial level of income. Theory predicts that low-income
countries tend to grow faster so that "convergence" in the level of
income among countries in the world will occur. Of course growth
depends on accumulation of labor, capital, and technology. Even adjusting for those factors, i.e., basically looking at TFP growth such as in
Tables 1 and 2, the TFP growth experience is impressive for most, not
all, of the eight countries.
I have reservations on this way of introducing a "miracle" and set up
the following analysis, as many papers do by now. As a student of "old
growth theory," I tend to distinguish a growth process, which is a
I am indebted to Anne O. Krueger and David Weil for their helpful comments on the
preceding version of this comments.
1. The World Bank (1993), The East Asian Miracle, has already been widely reviewed.
Rodrik (1993) challenges some of its assertions, especially links between hypotheses
and regression results. Amsden (1993) criticized the half-hearted appreciation of government interventions and industrial policy.
Comment 275
process of quantity changes, from economic development, which is
more about quality changes. My objection to a convergence regression
is that it ignores the nonlinear process, or qualitative change, of economic development.2
Simon Kuznets observed that when a country starts on a sustainable
growth path, qualitative changes have to occur. Investment grows
sharply, industrial structures change, and growth results with social
changes. Britain led the industrial revolution and modern economic
growth in the mid-18th century, followed by France, the United States,
and Germany in the first half of the 19th century. Many European
countries started rapid economic growth between 1860 and 1870, according to Kuznets. Kuznets (1959) coined the phrase "modern economic growth," listing some common features: (1) the application of
modern scientific thought and technology to industry; (2) a sustained
and rapid increase in real product per capita, usually (but not always)
accompanied by a high rate of population growth; (3) a rapid transformation of the industrial structure (changing sectoral output, labor force,
and capital stock distribution); and (4) an expansion of international
contacts. Kuznets, by the way, dates the start of modern economic
growth of Japan at around 1874 and 1879.
Was it also a "miracle" to have seen the European countries experience rapid economic growth around the turn of century? If Kuznets
dates modern economic growth for Asia in the year 2020, the date for
the Four Tigers would be sometime around 1960 and the ASEAN four
around 1970. Wouldn't it be more heuristic to compare East Asian
experiences not with the current developing nations but with the
European industrial revolution? Of course, Korea in 1960 faced quite
different economic conditions from Britain in 1780. However, the kind
of change in industrial structures, work habits, saving behavior, and
other social changes may be similar.
This is more than just a philosophical point. Consider the Solow
diagram of (Y/L) on (K/L). Depending on the shape of the production
function, there may be many steady states. Let us suppose that there
are two stable equilibria, the low, k*, and the high, k**, with an
intermediate, unstable steady state, kc. If a country can make a jump (in
a matter of a decade) from k* to somewhere above kc, the critical value,
then the "convergence" logic works. But an interesting part is not the
convergence process between kC and k**, but an initial jump from k*
to above kc. Is the jump a miracle or a big push?
2. The following argument is a summary of my previous comments on a convergence
paper by Easterly (1994); see Ito (1994).
276 ?ITO
I consider that it is more interesting to see what causes the jump from
a low equilibrium (vicious circle) onto a path toward a high equilibrium.
Once high investment kicks in, growth results, and growth fosters
confidence in the country and high savings result, which connects to
investment, a virtuous circle.3
Many countries made a jump or a takeoff to a high-growth path in
different times. We may be witnessing a group of countries taking off
from Asia. A benchmark to be compared should be the earlier development experiences by European countries, the United States, Canada,
Australia, and Japan in different times, rather than contemporary developing countries.
If this version of Kuznets's view of the world is accepted, then the
regular convergence regression does not make sense. At best, it is
misspecified in that it mixes up three groups of countries: one large
group of countries that are yet to take off-low level of incomes with
low rates of growth; a group of OECD countries-a
high level of
income with a low rate of growth; and a small group of countries that
rates with a low- to
are in the process of catching up-high-growth
middle-income level, i.e., "converging." What a convergence regression
attempts to capture is the third group of countries.
2. CommonDenominatorsof Success
The paper lists common denominators among the eight countries. First,
the paper observes that the growth experience was accompanied by
relative income equity. Then, four keys for success are explained: (1)
maintaining macroeconomic stability, (2) broad-based educational
strategies, (3) export growth, and (4) insignificance of industrial policy.
The first two are called "conventional," while the latter two are called
"unconventional." Obviously, few disagree with the conventional wisdom, and I make only passing remarks. I will make more extensive
comments on the unconventional ones, (3) and (4).
It is obvious that macroeconomic stability is important for growth.
But macroeconomic stability is easy to achieve when growth is high, a
possibility of reverse causality.4 When income is growing faster than
3. This is a simple argument of low- and high-income multiple equilibria in the framework of the Solow model. I certainly do not claim this argument to be particularly
original. Professor James Tobin pointed out to me during the conference that similar
points were made by Richard R. Nelson (1956, 1960).
4. In 1960, Japanese political power was paralyzed over the new security treaty with the
United States. A new 10-year plan "to double income" was proposed by a new prime
minister, Hayato Ikeda, to heal the wounds between the political right and left. Growth
as a national target was used to achieve political and macroeconomic stability. See Ito
(1992, pp. 61-67) for macroeconomic planning.
Comment*277
expected, tax revenue will be higher to help fiscal balance, and inflation
is easier to control.
Education is important. Usually, primary and secondary enrollment
rates or public expenditure on education, as in Tables 7 and 8 of this
paper, are used to show the evidence on high human capital accumulation.5 These variables are admittedly proxies. What is learned in an
education system, from knowledge accumulation to group conducts, as
well as education at home, may be more important in growth. But they
are hard to measure.
3. Sequencingas IndustrialPolicy
The paper argues that export push was a successful strategy, while
industrial policy was "insignificant." Exports have enhanced growth,
while trying to pick particular industries has not been successful. The
former assertion was supported by high correlations between some
measures of openness and exports, including the admittedly controversial Dollar index, and growth. The latter assertion was supported
mainly by evidence that TFP growth in promoted sectors in Japan,
Korea, and Taiwan were not particularly high, and promotion of particular industries did not change overall growth, compared with what it
would have been without intervention.
In discussing industrial policy, I will challenge the paper from two
aspects: What is important in industrial policy is "sequencing," which is
not discussed in the paper.
Sequencing of the leading industries in the economy is considered to
be a key to growth, according to an oral tradition of East Asian
countries.6 For example, industrial policy in Japan has been centered
around promoting industry after industry: textiles and toys, steel, chemicals, shipbuildings, to high-tech industries. Due to required sophistication in technology and large fixed costs, the government was aware of
the importance of sequencing industries. The developed "light and
low-tech" industries make a platform for a jump to the next stage,
"mid-tech" industries. A jump to "high tech" is only appropriate after a
5. Rodrik(1993) pointed out that growth can be explained by the "initial"condition of
high education levels and equity in income and land distributions,without using the
investment/GDP ratio that Rodrikcorrectlypoints out as "endogenous."
6. A broadconcept of sequencingin economicdevelopmentis often called a "flying geese
pattern"in the Japaneseliterature.Originally,the flying geese patternwas named after
the patternof importsubstitutionto export substitution,in one industry after another.
Later,it is directed to a pattern of regional countries' development sequencing; the
leader, Japan,is followed by Korea,Taiwan,Hong Kong, and Singapore,and they in
turn are followed by other ASEANcountries.
278 ?ITO
Japan
Textile
Clothing,
apparel
Korea
Taiwan Hong Kong
Singapore
1900-1930
Again 1950
1960s Early1950s Early1960s
1970s (dominant) Again 1970s
1950s
1960s 1950s-1960s
1970s
1960s 1960s-1970s
1970s
Toys, watches,
footwares
Refining
1950s
Steel
1950s-1960s
Chemicals
1960s-1970s
Shipbuilding
Electronics
1960s-1970s
1970s
Automobile
1970s-1980s
1980s
Computers,
semiconductors
Bankingand
finance
(promo)
Early1960s
(promo)
Latye 1960sEarly1970s
Late 1960s
1970s
Early1970s
1970s
Late 1970s
1980s
1980s
1980s
Late 1980s
1970s
Late 1970s
1980s
1980s
Source:Japan, Korea, Taiwan: author's judgment; Hong Kong, Singapore, Young (1992).
success at the mid tech" industries. Profits and experiences-both
management and labor-in a lower stage are key for a success in
developing a higher stage. Too high a jump to skip a stage, just like an
attempt to take off an airplane with too steep an angle, would result in
a crashing failure. East Asian countries, having observed a history of
European and U.S. economic history, have repeated the same process,
but faster.7
The table that appears in this Comment shows an experience among
East Asian countries of such sequencing.
If sequencing is important, then four implications follow. First, it is
not a "miracle" to observe takeoff in Asia. It is just repeating what
Europe did a century ago. (This is also pointed out in the first part of
these comments.) Second, when repeating a history, it is easy to predict
which industry comes next, hence "picking a sunrise industry" is not as
7. See Ito (1992, Chapter 2) for elaboration on economic takeoff and sequencing. Ito (1992,
Chapter 2, footnote 18) invokes an analogy to a biological idea that "ontogeny
recapitulates phylogeny."
Comment *279
risky as it is often claimed. Many Japanese economists argue that
promoting particular industries in the 1950s and 1960s was theoretically
possible and successful, but it became impossible and undesirable to
pick winners in the 1980s, as the economy became more "mature."
Third, showing low TFPs in "promoted sectors" cannot be used as
evidence for a failure of industrial policy. If industrial policy made it
possible to run up the stages faster, it is a success even without realizing
high TFPs in each stage.8
Fourth, showing that the composition of industries in Japan and
Korea is not different from the theoretical prediction (Table 12 of Page's
paper) cannot be taken as evidence of the irrelevance of industrial
policy. Even if the composition after intervention is the same as the
composition predicted from theory, the success is getting there fast.
4. RegionalSecret?
However, the puzzle still remains. If an initial push can be done
anywhere in the world, one should observe high growth taking place
randomly on the globe. This is not the case. The 19th-century growth
was concentrated in Europe and North America (with an exception of
Australia and Japan), while the late 20th-century growth is happening
in Asia. Certainly geographical proximity plays an important role.
Technological transfers, worker mobility, trade links, and political peer
pressure may be important factors in regional development. It happened in Asia in the 1980s. Maybe another "miracle" will occur somewhere else in the 2000s.
REFERENCES
Amsden, A. H. (1993).Why isn't the whole world experimentingwith the East
Asian model to develop? Review of the World Bank's East Asian Miracle
Report,to appear in Symposium on the World Bank's East Asian Miracle
Report, in World Development.
Easterly, W. (1994). Explaining miracles: Growth regressions meet the gang of
four. In NBER, East Asian seminar on economics1993, T. Ito and A. O. Krueger
(eds.).
Ito, T. (1992). The Japaneseeconomy. Cambridge, MA: The MIT Press.
(1994). Comments on William Easterly, Explaining miracles: Growth
regressions meet the gang of four. In NBER, East Asian seminar on economics
1993, T. Ito and A. O. Krueger(eds.).
Kuznets, S. (1959). Six lectures on economicgrowth. Glencoe, IL: Free Press.
8. The point made by Young (1992),in that Singaporeinvested too fast to enjoy learning
curve effects,is also not taken as a failurefrom the sequencing point of view.
280 ?DISCUSSION
Nelson, R. R. (1956).A theory of the low-level equilibriumtrap in underdeveloped economies. AmericanEconomicReview46:896-908.
(1960). Growth models and the escape from the low-level equilibrium
and CulturalChange8:378-388.
trap: The case of Japan. EconomicDevelopment
Rodrik, D. (1993). King Kong meets Godzilla: The World Bank and The East
AsianMiracle.Mimeo, December.
Young, A. (1992).A tale of two cities:Factoraccumulationand technicalchange
in Hong Kong and Singapore. NBERmacroeconomic
annual1992. Cambridge,
MA:The MITPress.
Discussion
Page responded to several points raised by Collins and by Ito. He
agreed with Collins that creating a dichotomy between the productivity
growth and accumulation views was misleading. The more important
question, suggested Page, is how these economies performed so well on
both fronts and how they were able to generate and sustain such high
rates of accumulation of physical and human capital without the efficiency losses observed in the former Soviet Union and in Eastern
Europe. Page also agreed with Collins that it is important to address the
political economy issue of policy implementation and to understand the
particular institutions that might explain the success of the macroeconomic managers and bureaucracies in these countries.
Page clarified the distinction made in the paper between industrial
policy and export promotion. According to the export promotion view,
the actual process of exporting is special, rather than the goods that are
produced. Under the industrial policy view, however, it is the particular
goods and industries that are important. Page agreed that it is often
difficult to disentangle the effects of the two stories, especially since the
criteria for success of industrial policies often seems to be related to
export performance. While Page felt that the results of the paper and
other more micro-oriented studies support the export-promotion story,
he noted that more case-study evidence was needed before reaching
stronger conclusions. Page also suggested that one way to interpret
industrial policy in Japan and Korea was as a response to coordination
failures or market imperfections. He noted that this would make it more
difficult to determine whether industrial policy was successful since
such a policy would appear market-conforming, ex post.
Olivier Blanchard questioned the use of the high-income country
elasticity for the computation of TFP growth. The standard approach is
either to use capital's share of income as the elasticity or to estimate the
Discussion?281
elasticity. Page responded that data on capital income shares were not
available. In estimating the elasticities, the issue is how to separate the
effect of TFP growth from the effects of allocative inefficiencies. For a
subset of developing countries in the sample, output growth was
negative, while factor accumulation was positive, suggesting severe
resource allocation problems that would introduce noise in the production function estimates. Therefore, the elasticities were computed from
the high-income countries since they were the least likely to suffer
these allocation inefficiencies.
Herschel Grossman commented that an important prior question is
whether initial conditions allowed particular countries to achieve highgrowth rates. If certain growth rates are economically infeasible given
the available resources, then the political economy issue of how institutions foster growth is a secondary issue. Page responded that the initial
conditions that are imposed by the data are presumably affected by
prior public policy actions, which is why understanding the evolution
of institutions is important. He added that the historical record on
identifying necessary conditions for growth and predicting growth
rates was poor. For example, in the late 1950s the USAID identified
Burma and the Philippines as the growth poles for East Asia and wrote
that the most serious impediment to South Korea's economic development was its bureaucracy.
Andrew Atkeson asked whether it was exports or trade more generally that was connected to growth. He noted that factor proportions
theory would suggest that by opening up to trade, the economy could
buy more capital services or intermediate inputs. This might be related
to the sequencing theory advanced by Ito, if factor proportion differences were viewed as the reason that the economy moved from industry to industry. Page answered that the high trade to GNP ratios for the
HPAEs were not achieved by incentive-neutral policies, but rather by
highly selective promotion of exports and admission of imports. He
added that there was evidence from the micro literature that firms
competing with imports and firms that compete on the export market
but within the same industry have different learning effects.
John Cochrane commented that important variables in the regressions such as , vings, education, and growth, were endogenous, and
questioned whether any causal statements could be made. Cochrane
interpreted the experience from the Soviet Union and Eastern Europe
as evidence that an exogenously high savings rate doesn't cause growth,
suggesting that the observed positive correlation between savings and
growth should be interpreted as the result of countries saving because
they know that their growth prospects are good.